GEORGIA INSTITUTE OF TECHNOLOGY 

Under the provisions of the regulations for the degree 

MASTER OF INDUSTRIAL DESIGN 

on

Friday, May 1, 2026

10:00 a.m. – 11:00 a.m. EST 

West Architecture 155

Teams Link: Mingqi Wang Thesis Defense | Meeting-Join | Microsoft Teams

Meeting ID: 294 048 181 186 72

Passcode: WV6Mj2YH

Mingqi Wang

will present a thesis defense entitled, 

"Designing for Breathing-Stroke Coordination Awareness and Training in Front Crawl Swimming"

 Advisor: 

Dr. Yixiao Wang, Georgia Tech School of Industrial Design 

Committee:  

Dr. Eunsook Kwon, Georgia Tech School of Industrial Design 

Dr. Alexander T. Adams, Georgia Tech School of Interactive Computing

Faculty and students are invited to attend this presentation. 

Abstract  

Breathing-stroke coordination in front crawl swimming is a skill most swimmers cannot directly perceive or correct in real time. Every breath requires precise lateral head rotation timed to body roll and the arm pull cycle, yet swimmers have no direct view of their own mechanics while in motion. Traditional coaching addresses this through delayed verbal feedback, and while commercial swimming technology has expanded, no existing product targets breathing-stroke coordination specifically.

This thesis investigates how technology can support swimmers' awareness of breathing-stroke coordination in front crawl, with a focus on recreational and developing swimmers. The research follows a two-phase design process. Phase 1 consisted of semi-structured interviews with 16 swimmers and coaches spanning beginner through elite levels, analyzed through thematic analysis to characterize how coordination is experienced across skill levels and to derive design guidelines. Phase 2 developed a virtual reality training prototype grounded in these guidelines, presenting synchronized hand trajectory visualizations and spatial audio cues to make the timing relationship between stroke mechanics and breathing windows explicitly visible. A feasibility pilot study with seven amateur swimmers was conducted to explore the prototype's usability and its potential as a coordination training tool.

Student Name: Xinqiang Rao

Thesis Title: Multiscale Computational Modeling of Defect-Controlled Degradation in Layered Cathode Materials

Thesis Advisor: Dr. Julia H. Yang

Thesis Co-Advisor: N/A

Committee Members: Dr. Nian Liu, ChBE; Dr. Andrew J. Medford, ChBE; Dr. Solomon Tolulope Oyakhire, ChBE; Dr. Seung Soon Jang, MSE

Date: 5/5/2026

Time: 1:00 to 3:00 PM ET

Location: Ford ES&T Conference Room 2229

Kaely Hall

Ph.D. Student in Human-centered Computing 

School of Interactive Computing 

Georgia Institute of Technology 

Date: Friday, May 8th, 2026

Time:  12-3pm

Location: Coda 1215 Midtown  

Teams Info: https://teams.microsoft.com/meet/2255456448249?p=rIzXwFXWGRftAe5rTE 

Meeting ID: 225 545 644 824 9

Passcode: X7zT7AR3

Committee 

Dr. Jennifer Kim (advisor) - School of Interactive Computing, Georgia Institute of Technology

Dr. Munmun de Choudhury - School of Interactive Computing, Georgia Institute of Technology

Dr. Nasibeh Farahani - Mayo Clinic Platform, Mayo Clinic

Dr. Andrea Parker - School of Interactive Computing, Georgia Institute of Technology

Dr. Vedant Das Swain - Tanden School of Engineering, New York University

Abstract 

Artificial intelligence (AI) systems are increasingly used to support how individuals construct self-representations in everyday and decision-critical contexts. For example, people use AI to draft professional materials such as cover letters and bios, or to articulate preferences in medical settings, such as birth plans that communicate values and priorities for care. In these contexts, AI use extends beyond task assistance and becomes part of the process through which individuals construct and communicate representations of themselves. However, these representations often reflect the assumptions, conventions, and statistical patterns embedded in AI systems, resulting in outputs that may not faithfully capture users’ situated experiences, self-understandings, or how they intend to be understood. This dissertation introduces representational alignment as a framework for understanding this challenge: the extent to which AI-mediated representations preserve and convey a person’s intended meaning across contexts. Misalignment arises when a person’s intended self-representation diverges from the representation constructed by an AI system, often due to generalized or decontextualized interpretations of user input.

In my completed work, I examine representational alignment in two stages: breakdown and repair. The first study analyzes how misalignment emerges in practice through neurodivergent job-seekers’ use of an LLM-powered career support chatbot, showing that systems frequently misinterpret implicit cues and impose normative language and assumptions. As a result, the chatbot’s “support” is often grounded in misaligned representations of users’ experiences and intentions. The second study examines how users work to restore alignment through interaction. I introduce bi-directional alignment as an interaction paradigm in which users and systems iteratively shape representations together, and investigate this concept through LL.me, a research probe that supports iterative refinement of AI-generated professional self-representations. Findings show that alignment develops through reflection, reinterpretation, and refinement, as users incorporate tacit contextual knowledge, personal values, and anticipated audience expectations.

My proposed work extends representational alignment to clinical settings, where individuals must construct self-representations through the articulation of expectations preferences without full visibility into how they will be interpreted or executed within care institutions. Birth planning is a particularly challenging case, as patients must express delivery preferences in advance of uncertain and evolving conditions, where those preferences may later be reinterpreted, constrained, or overridden during care. I propose AIDoula, an interactive system designed to support patients in articulating and refining birth preferences by making clinical constraints, such as standard delivery intervention protocols, more visible and enabling iterative revision. In this context, AI-generated representations must function as boundary objects, carrying meaning between patients and clinicians while remaining actionable in care settings. Through a two-phase evaluation with patients and clinicians, I will examine how individuals construct conditional and flexible representations under partial constraint visibility, and how these representations are interpreted and reshaped within clinical workflows. Collectively, this dissertation advances a framework for representational alignment in human–AI systems, contributing empirical evidence of misalignment, interactional approaches for supporting alignment in practice, and a system that demonstrates how representations can remain meaningful under uncertain and evolving constraints.

Doctor of Philosophy in Ocean Science and Engineering

In the School of Civil and Environmental Engineering

Savannah L. Howard

Will defend her dissertation

FUNDAMENTAL HYDROGEOMECHANICAL PROPERTIES OF CARBONATE FORMATIONS: IMPLICATIONS FOR OFFSHORE COMPRESSED AIR ENERGY STORAGE

May 5th, 2026 at 8 AM

Ford ES&T, Room 3243 (The Ocean Room)

Teams: https://teams.microsoft.com/meet/252549462205277?p=eDS8aBGnt00SviVjBt

Meeting ID: 252 549 462 205 277

Passcode: hx7Vz7zX

Thesis Advisors:

Dr. Sheng Dai

School of Civil and Environmental Engineering

Georgia Institute of Technology

Committee Members:

Dr. Philipp Braun

Navier Laboratory

Ecole des Ponts, Paris Tech (ENPC)

Dr. Kevin Haas

School of Civil and Environmental Engineering

Georgia Institute of Technology

Dr. Joseph Montoya

School of Biological Sciences

Georgia Institute of Technology

Dr. Zhigang Peng

School of Earth and Atmospheric Sciences

Georgia Institute of Technology

Summary:

The U.S. energy grid faces a critical storage challenge: renewable sources such as wind and solar generate electricity intermittently, while demand peaks occur independently of generation. Among available grid-scale storage technologies, Compressed Air Energy Storage (CAES) — which stores energy by compressing air underground and releases it through turbines on demand — offers the lowest levelized cost of storage of any current technology, surpassing batteries, pumped hydro, hydrogen, and thermal systems. Existing CAES plants (Huntorf, Germany; McIntosh, Alabama) use salt caverns, which are geographically restricted. Porous carbonate formations, distributed across the southeastern U.S. coastal zones where offshore wind and wave energy resources are abundant, offer a far more geographically widespread alternative — but their suitability as CAES reservoirs under realistic subsurface conditions has never been systematically characterized.

This dissertation presents a comprehensive laboratory characterization of three distinct carbonate formations - St. Maximin, Miami, and Hemingway. Experiments characterized the full suite of properties relevant to CAES viability, including mineralogy (XRD), density, total and accessible porosity, micro-CT pore imaging, permeability, ultrasonic wave speeds and elastic moduli, unconfined compressive strength and failure mode (via acoustic emissions), water retention curves, creep behavior, and cyclic loading response. Energy storage density was calculated directly from measured pore and mechanical properties and benchmarked against the two operating CAES plants. The results highlight that pore geometry and connectivity must be characterized jointly with porosity for meaningful CAES modeling; water retention and evaporation govern energy storage density; creep is saturation-dependent and operationally significant; and cyclic pressurization does not cause rapid structural and strength deterioration, but cumulative strain can lead to failure in long-term operation. Calculated energy storage densities for offshore carbonate CAES are competitive with, and in high-porosity cases exceed, those of existing salt cavern plants (~7.5–18 MJ/m³), demonstrating the technical feasibility of porous carbonate formations as CAES reservoirs. The work establishes the first systematic experimental dataset for air–water saturated carbonate reservoir media under CAES-relevant conditions, filling a critical gap in a field previously dominated by oil–water and CO₂ storage studies. Findings directly inform site selection criteria, reservoir performance modeling, and operational safety assessments for future offshore CAES development in the southeastern U.S. — a region that combines abundant offshore renewable energy resources with extensive shallow carbonate geology.

Date: Monday, April 27th, 2026

Time: 1:00–3:00 PM ET

Location: online [Teams link]

Kaige Xie

Ph.D. Student in Computer Science

School of Interactive Computing

Georgia Institute of Technology

Committee:

Dr. Pascal Van Hentenryck (advisor) - School of Industrial and Systems Engineering and School of Interactive Computing, Georgia Institute of Technology

Dr. Thomas Ploetz - School of Interactive Computing, Georgia Institute of Technology

Dr. Chao Zhang - School of Computational Science and Engineering, Georgia Institute of Technology

Abstract:

This dissertation investigates how to optimize natural language generation (NLG) systems built on large language models (LLMs) from a holistic, lifecycle-oriented perspective. While recent advances in LLMs have led to substantial gains across a wide range of NLG tasks, prior research has largely focused on improving benchmark performance, often overlooking the broader challenges that arise across model training, inference, evaluation, and deployment. This dissertation argues that such a performance-centric view is insufficient for real-world NLG systems, whose success depends not only on output quality but also on efficiency, reasoning capability, evaluation fidelity, and user trust. To address this gap, the dissertation introduces a unified framework centered on text sub-structures—semantically meaningful intermediate representations embedded in text—and studies how their recognition and strategic utilization can improve NLG systems throughout their full lifecycle. 

The dissertation develops this framework across four representative NLG tasks: dialogue summarization, story generation, action plan generation, and question answering. In dialogue summarization, it shows how dialogue skeletons can facilitate more effective few-shot learning and improve cross-task prompt transfer under limited supervision. In story generation, it demonstrates how outline-based planning structures can guide LLMs toward producing more coherent and engaging narratives. In action plan generation, it examines precondition-effect dependencies as a form of latent world knowledge that enables LLMs to better model action feasibility and environmental change. In question answering, it explores sub-questions as a versatile sub-structure for both fine-grained system evaluation and explanation generation, improving the assessment of open-ended retrieval-augmented generation systems and enhancing users’ ability to judge model reliability. Collectively, these studies show that text sub-structures provide a general and effective semantic scaffold for improving learning efficiency, inference-time planning and reasoning, evaluation robustness, and deployment-time user experience. 

Committee:

Dr. Ayazi, Advisor

Dr. Taylor, Chair

Dr. Ansari

Committee:

Dr. Gaylord, Advisor

Dr. Cai, Chair

Dr. Adibi

Date: Tuesday, April 28th, 2026
Time: 11:00 am – 1:00 pm EST
Location: Groseclose 226 (Georgia Freight Bureau Conference Room) and Zoom

Committee:
Dr. Santanu S. Dey (Advisor), H. Milton Stewart School of Industrial and Systems Engineering, Georgia Institute of Technology
Dr. Oktay Günlük, H. Milton Stewart School of Industrial and Systems Engineering, Georgia Institute of Technology
Dr. Jean-Philippe P. Richard, Department of Industrial and Systems Engineering, University of Minnesota
Dr. Nick Sahinidis, H. Milton Stewart School of Industrial and Systems Engineering, Georgia Institute of Technology
Dr. Yang Wang, School of Civil and Environmental Engineering, Georgia Institute of Technology

Abstract:
Quadratically constrained programs (QCQPs) model critical real-world problems across engineering, finance, logistics, and data science. For nonconvex QCQPs, finding globally optimal solutions is computationally intractable in general, and even small instances can be challenging for state-of-the-art optimization solvers. This thesis studies the design and analysis of global optimization algorithms and relaxation schemes for QCQPs that are both mathematically rigorous and computationally practical.

Solving QCQPs to global optimality relies on three key components: (i) finding high-quality feasible solutions, (ii) constructing tight relaxations, and (iii) selecting effective branching strategies. These are crucial for the spatial branch-and-bound, which is the dominant approach for solving QCQPs. This thesis contributes to each of these components.

The first part of the thesis addresses the problem of finding high-quality feasible solutions for an energy storage optimization problem whose operational constraints are bilinear. We develop a regularized formulation whose linear programming relaxation yields a feasible solution to the original problem. This approach enables solving previously challenging trilevel optimization problems modeling adversarial attacks on power grids.

The second part of the thesis focuses on building a tighter convex relaxation for bilinear bipartite programs using aggregation techniques to construct second-order cone representable sets. We characterize when aggregation methods yield the exact convex hull and provide illustrative examples. Computational experiments demonstrate that aggregation methods improve bounds over single-row relaxations and commercial solvers, reducing the optimality gap.

The third part of the thesis proposes a novel branching rule, namely extreme strong branching, which jointly optimizes both branching variable and point selection via binary search. This approach leverages objective value improvements to minimize branch-and-bound tree size while producing bound tightening as a byproduct. For certain instances, this technique produces substantially smaller branch-and-bound trees to reach optimality compared to alternative branching rules.

The last part of the thesis studies relaxation schemes in a unified framework, from data-independent to data-dependent approaches. We prove that bounded Lagrangian relaxation provides the strongest bound among data-independent methods under mild conditions. We further analyze data-dependent relaxations and establish a hierarchy of the relative strength of various relaxation methods. 

Collectively, these contributions bridge the gap between theoretical possibility and computational feasibility, expanding the range of QCQP instances that can be addressed effectively in practice.

Committee:

Dr. Chatterjee, Advisor

Dr. Hao, Chair

Dr. Mukhopadhyay

Advisor: Dr. Dimitri Mavris

Milestone: PhD Thesis Final Examination (Defense)

Degree Program: Aerospace Engineering

Title: An Approach to Developing Ontology-Based Epistemic Infrastructure for Early-Phase Architecting of Space Exploration Campaigns

Abstract: NASA's Artemis Program, an ambitious space exploration campaign (SEC) to explore the Moon and Mars, is expected to grow to over 50 highly-coupled major acquisitions. Yet for the past 35 years, NASA's acquisition practices have been designated "high-risk" by the GAO due to chronic cost overruns and schedule delays. A root cause analysis traces these outcomes to epistemic shortfalls in early-phase architecting: tacitness, ambiguity, manual data transfers, and inconsistencies in how architectural knowledge is represented, exchanged, and verified. To address these shortfalls, this dissertation introduces ontologies and semantic web technologies into early-phase campaign architecting to develop ontology-based epistemic infrastructure. A tailored methodology, SEC Ontology Development Methodology (SECODM), was formulated. Following it, a layered ontology ecosystem, SECO, was developed along with minimum viable infrastructure for development, verification, and deployment. An evaluation framework based on elegant systems theory was developed to assess these contributions. Three case studies demonstrated that the ontology and infrastructure address the identified root causes through formal knowledge representation, machine-readable interoperability, and automated consistency verification.

Date and time: 2026-05-04, 1pm ET

Location: CoVE

Committee:
Dr. Dimitri Mavris (advisor), School of Aerospace Engineering
Prof. Daniel Scrage, School of Aerospace Engineering
Prof. Thomas Gonzalez Roberts, School of Aerospace Engineering
Dr. Bradford Robertson, School of Aerospace Engineering
Dr. Michael Balchanos, School of Aerospace Engineering
Dr. Stephen Edwards, NASA MSFC

Advisor: Dr. Marilyn Smith

Milestone: PhD Thesis Proposal

Degree Program: Aerospace Engineering

Title: Reduced-Order Modeling for Aero-Structural Design of Advanced Air Mobility Vehicles

Abstract: In recent years, the field of Advanced Air Mobility (AAM)---previously referred to as Urban Air Mobility (UAM)---has made significant progress towards commercialization. These AAM vehicles have consumer-facing applications such as air taxi and Uncrewed Aerial Vehicle (UAV) delivery services, with many proposed missions occurring within urban settings. This is an environment characterized by aerodynamic drivers that are not present in the traditional flight analysis methods applied in open-air conditions, such as the importance of building wake flows and small-scale turbulence. The AAM industry is also presently facing the challenging economics of bringing financially stable low capacity flights to the market, meaning minimizing costs while maximizing safety are a large concern. With this in mind, a deeper understanding of sources of structural fatigue and their significance may be crucial to keeping inspection and repair costs low while maintaining confidence in vehicle and passenger safety. This thesis aims to elucidate the feasibility of reduced-order models as alternatives to costly mid- or high-fidelity computational tools for early design and analysis of AAM's. Specifically, aerodynamic flow models will be assesed for their sensitivity and applicability in urban environment applications. Addionally, the effects of multi-tiltrotor downwash on a wing will be evaluated with a number of solvers. The resulting structural loads will be compared to determine the bounds of the tradeoff between model fidelity and solution accuracy.

Date and time: 2026-04-29, 2:00pm

Location: Skiles 246

Committee:
Dr. Marilyn Smith (advisor), School of Aerospace Engineering
Dr. Juergen Rauleder, School of Aerospace Engineering
Dr. Graeme Kennedy, School of Aerospace Engineering
, 

 

Advisor: Dr. Timothy Lieuwen

Milestone: PhD Thesis Proposal

Degree Program: Aerospace Engineering

Title: Unsteady Wake Structures in Jets in Crossflow

Abstract: The jet in crossflow (JICF) is a fundamental, fluid mechanics-rich flow topology employed in various combustor architectures such as lean premixed combustors, staged combustors, as well as in film cooling technologies, owing to its excellent mixing properties. While on paper the implementation of this flow configuration is simple enough, the interaction between the two fluid streams results in a super three-dimensional, unsteady large scale vortical structures, namely: horseshoe vortices, shear layer vortices, counter-rotating vortex pairs, and the wake vortices. Out of these, the wake vortices are the least understood and have received limited attention. A large body of literature exists in characterizing the formation and development of the shear layer vortices (SLV) and the counter-rotating vortex pairs (CVP), and as such, control strategies are developed around their behavior in system level applications. Despite the bulk of the mixing is attributed to the SLVs and the CVP, the wake structures appear to add another layer of far field mixing from their tornado-like behavior and are the major focus of this work. Although experimental and computational studies have been performed to characterize the periodicity associated with the wake vortices, general consensus on their precise formation mechanism and source of vorticity is yet to be achieved. Moreover, despite the existence of tornado-like structures in the wake of the jet are widely acknowledged in the research community, the exact nature of their evolution is yet to be explained. In the context of a reacting jet in crossflow (staged combustors, for instance), presence of combustion has been shown to significantly alter the underlying hydrodynamics of the flow field due to additional physics like dilatation effects, baroclinic torque, and local viscosity changes, resulting in complex flame-flow interactions. Furthermore, combustor phenomena such as flame stabilization and Nox emissions have been shown to be highly sensitive to the background fluid mechanics of this flow field, necessitating a systematic investigation onto combustion effects. Additionally, since the evolution of these large scale flow structures occurs at their respective time and length scales, studies on their interactions require highly-resolved spatial and temporal datasets. Motivated by the above-mentioned gaps in literature and challenges on analysis, this proposal seeks to explore the characteristics of the wake structures and attempts to answer fundamental questions like: (1) What is the nature of flow separation mechanisms (low pressure-driven events vs CVP induction) and source of vorticity (crossflow boundary layer vs jet shear layer), (2) what is the effect of flame attachment location and the consequent local heat release distribution on wake vortices, (3) what are the unsteady features of the wake vortices (characteristic frequencies, symmetry, mode shapes) and how do they compare between non-reacting and reacting jet in crossflow datasets. These research questions are systematically investigated using high fidelity large eddy simulations. The computations will be performed over a range of fluid dynamic parameters characterizing the flow field and the dependencies of the topological features of wake structures on test parameters will be ascertained. 

Date and time: 2026-05-01, 9.00 AM - 11.00 AM

Location: MK-317

Committee:
Dr. Timothy Lieuwen (advisor), School of Aerospace Engineering
Dr. Ari Glezer, School of Mechanical Engineering
Dr. Joseph Oefelein, School of Aerospace Engineering
, 
, 
, 

Ayush Dhande

will defend Master's thesis entitled,

Development of Intrinsically Stretchable Vertical Organic Electrochemical Transistors via Interfacial Top Electrode Modification

On

Tuesday, April 21 at 3:00 pm
 MRDC 4404

And

Virtually via

https://teams.microsoft.com/meet/265405960763671?p=3K2IxWxSKMV9gb9ZJc

Committee
            Prof. Antonio Facchetti – School of Materials Science and Engineering (Advisor)
            Prof. Jason Azoulay – School of  Materials Science and Engineering
            Prof. Seung Soon Jang – School of Materials Science and Engineering

Abstract

Organic electrochemical transistors (OECTs) have emerged a promising technology for bio-integrated electronics due to their low voltage operation and high transconductance, yet achieving both performance and mechanical stretchability remains challenge. In this work, an intrinsically stretchable vertical OECT (svOECTs) is fabricated to bridge the gap, utilizing a vertically stacked architecture that enables high device density and nanometers scale channel lengths. To address the failure of electrodes under mechanical strain, a top electrode engineering strategy is introduced using a trilayer top electrode mechanism. This significantly enhances interfacial adhesion and charge transport. The resulting svOECTs exhibit electromechanical stability, durability and maintain functionality under different mechanical conditions. This study demonstrates that combining vertical device architecture with interfacial molecular engineering provides a framework for next generation bioelectronics of high fidelity sensing on dynamic biological surfaces.

Student Name: Jacob Alan DeWitt

Thesis Title: Mechanocatalytic Synthesis of Ammonia over Transition Metal Nitrides

Thesis Advisor: Dr. Carsten Sievers

Thesis Co-Advisor: N/A

Committee Members: Dr. William J. Koros (School of Chemical & Biomolecular Engineering), Dr. David W. Flaherty (School of Chemical & Biomolecular Engineering), Dr. Faisal M. Alamgir (School of Materials Science and Engineering), Dr. Justin Hargreaves (School of Chemistry - University of Glasgow)

Date: 05/01/2026

Time: 9:00 am EST

Location: RBI 114

BioE Ph.D. Defense Presentation

Wed, May 6, 2026, 12:30 PM

Location: EBB CHOA Seminar Room

https://gatech.zoom.us/j/96972528891?pwd=wfP5hwf2uLElzalXRKe0aCtPJxomYc.1&from=addon

 Advisor: Todd Sulchek, Ph.D. (ME, Georgia Institute of Technology)

Committee:

Alexander Alexeev, Ph.D. (ME, Georgia Institute of Technology)

Scott Danielsen, Ph.D. (MSE, Georgia Institute of Technology)

David Myers, Ph.D. (BME, Georgia Institute of Technology)

Guillem Pratx, Ph.D. (Radiology and Medical Physics, Stanford University) 

Optimization of Forces, Loading Rates and Strain for Cytosolic Delivery using Mechanoporation  

Efficient intracellular delivery of cargo remains a central challenge in biomedical engineering to improve gene editing, cell therapy production, and cell diagnostics. The optimal delivery not only introduces cargo into cells but achieves cytosolic access while preserving cell viability and normal function. Existing approaches often fall short of this balance. Viral and carrier-mediated methods can promote uptake of molecularly encoded genes but are limited by endosomal entrapment and intracellular degradation. Electroporation can disrupt cell membranes but also disturbs normal cellular physiology. These limitations have motivated interest in microfluidic mechanoporation, in which cells are transiently permeabilized through controlled mechanical deformation. However, the mechanism of mechanoporation remains poorly defined and, as such, performance remains difficult to predict and standardize. To bridge the gap, this dissertation establishes a framework connecting device geometry, transient loading dynamics, and delivery outcomes. A key learning is that mechanoporation performance is demonstrated to be better explained by cell loading kinetics than by deformation magnitude alone. Systematic channel width and ridge angle variation demonstrates how tuning force rate, directionality, and loading history improves delivery efficiency, cell survival and experimental consistency. Integration of high-speed trajectory analysis, computational fluid dynamics (CFD), and multivariate modeling further reveals that steady-state descriptors such as strain magnitude are insufficient to explain mechanoporation outcomes, which instead, correlate strongly with strain rate and history-dependent hydrodynamic forces. Extending these insights to nanoparticle delivery, imaging analyses demonstrate that optimized device geometries enable deeper intracellular access and improved cargo retention relative to conventional incubation-based approaches. Together, these results provide a force-based framework for mechanoporation that links device design to biologically meaningful intracellular delivery.

BioE PhD Defense Presentation- Avi Gupta

Summarize

Paige, Laura T<laura.paige@bioengineering.gatech.edu>

 

​Paige, Laura T​

​Nelson, Teresa L;​announcements@grad.gatech.edu​

Avi Gupta

BioE Ph.D. Defense Presentation

Wed, May 6, 2026, 12:30 PM

Location: EBB CHOA Seminar Room

https://gatech.zoom.us/j/96972528891?pwd=wfP5hwf2uLElzalXRKe0aCtPJxomYc.1&from=addon

 

 Advisor: Todd Sulchek, Ph.D. (ME, Georgia Institute of Technology)

 

Committee:

Alexander Alexeev, Ph.D. (ME, Georgia Institute of Technology)

Scott Danielsen, Ph.D. (MSE, Georgia Institute of Technology)

David Myers, Ph.D. (BME, Georgia Institute of Technology)

Guillem Pratx, Ph.D. (Radiology and Medical Physics, Stanford University) 

 

Optimization of Forces, Loading Rates and Strain for Cytosolic Delivery using Mechanoporation  

]]>

Bioengineering PhD Defense Presentation

April 29, 2026 (Wednesday)

10:00 AM

Location: Howey N201/202

Meeting link: https://gatech.zoom.us/j/91549406195?pwd=ob1cyXlVnDSpDTjJqx6hQBqzIhnbVz.1

Advisor:  

James C. Gumbart, Ph.D. (School of Physics, Georgia Institute of Technology)  

Committee:  

M. G. Finn, Ph.D. (School of Chemistry and Biochemistry, Georgia Institute of Technology)

Peter Kasson, Ph.D. (School of Chemistry and Biochemistry; Coulter Department of Biomedical Engineering, Georgia Institute of Technology)

Jeffrey Skolnick, Ph.D. (School of Biological Sciences, Georgia Institute of Technology)

Adegboyega Oyelere, Ph.D. (School of Chemistry and Biochemistry, Georgia Institute of Technology)

Modulating Hepatitis B Virus Capsid Assembly and Function through Pharmacological Intervention  

Hepatitis B virus (HBV) is a major global health burden that causes chronic liver disease. Its replication depends on the assembly of a protein capsid, making this process an important target for antiviral intervention. This work integrates molecular dynamics simulations and biophysical experiments to investigate how small molecules interact with and modulate the HBV core protein across different structural contexts. First, biophysics-guided approaches were used to identify ligands targeting the inter-dimer assembly interface, showing that small molecules can perturb capsid assembly through changes in inter-subunit interactions. Second, free-energy calculations were conducted to characterize the conformational landscape of a key assembly intermediate, providing insight into structural features that may influence assembly nucleation. Third, a distinct intra-dimer pocket was characterized using fragment-based mapping and pharmacophore modeling, enabling virtual screening and experimental validation of ligands with extended binding modes. Together, these results provide a framework for understanding how small molecules can modulate HBV capsid behavior and highlight new directions for antiviral design.

Title: Tensor-based Predictive Modeling and Control for High-dimensional Data

Advisor: Dr. Jonathan Rogers

Milestone: MS Thesis Proposal

Degree Program: Aerospace Engineering

Title: UAV Implementation of Optimal Path Planning for Radiological Source Localization in Cluttered Environments

Abstract: This proposal presents an algorithm to optimize a radiological search path by leveraging a path-length optimization method combined with a measure of total information gain. A constrained optimization problem is solved in which the goal is to select measurement locations that yield the best-conditioned estimation problem, with the constraint that the total travel time to visit all points must be less than a specified value. The effectiveness of search locations is measured by the stable rank of the kernel matrix, which quantifies the attenuation between each candidate source location and candidate measurement location. Measurement locations are selected by a greedy algorithm for maximum stable rank while the total travel time is minimized by solving the well-known traveling salesman problem. Simulation results of the algorithm demonstrate that the optimization improves the estimation of radiological source term parameters for a given time constraint. The continued work will focus on the implementation of the algorithm on an unmanned aerial vehicle (UAV). Modifications to the algorithm will be needed to account for different absorption mediums and isotopes. A lab-based experiment demonstrating a search for radiation point sources of different isotopes is being designed.

Date and time: 2026-04-24, 3:00-4:00

Location: Weber 200

Committee:
Dr. Jonathan Rogers (advisor), School of Aerospace Engineering
Dr. Lu Gan, School of Aerospace Engineering
Dr. Craig Bakker, Pacific Northwest National Laboratory
,

BioE Ph.D. Defense Presentation

Fri, May 1, 2026, 12:00 PM – 2:00 PM

Location: IBB 1128 (Suddath Seminar Room)

https://gatech.zoom.us/j/2361182737?pwd=a2N4aFVRZzlHcXUwazBDdGROMUhoUT09

Advisor:

Ahmet F. Coskun, Ph.D. (Biomedical Engineering, Georgia Institute of Technology)

Committee:

Peng Qiu, Ph.D. (Department of Biomedical Engineering, Georgia Institute of Technology)

Saurabh Sinha, Ph.D. (Department of Biomedical Engineering, Georgia Institute of Technology)

Rabindra Tirouvanziam, Ph.D. (Department of Biomedical Engineering, Georgia Institute of Technology)

Marcus Cicerone, Ph.D. (Department of Chemistry & Biochemistry, Georgia Institute of Technology)

Timeseries Spatial Omics of Immune Cell Signaling in Inflammation

Cystic fibrosis (CF) affects over 162,000 people worldwide. Although FDA-approved therapies have shown promising results, chronic inflammation and bacterial infections persist, causing tissue damage and worsening quality of life despite a large immune cell presence. This thesis investigates the spatiotemporal dynamics of NF-κB signaling during inflammation by developing and applying spatial omics tools at the single-cell level across multiple biological contexts.

We first introduce PRISMS, an open-sourced, automated multiplexing pipeline for spatial transcriptomic and proteomic imaging compatible with Nikon widefield and Cephla spinning disk confocal microscopy. Using PRISMS, we apply pSigOmics static fixation to profile over 100,000 mouse fibroblasts stimulated with TNFα and IL-1β, revealing a novel asymmetric protein-RNA (APR) relationship between p65 RNA and protein. Graph neural network classification and PHATe trajectory analysis further delineate distinct APR subpopulations among stimulated fibroblasts.

We next extend pSigOmics with programmable fixation via helical perfusion to create a continuous gradient of cell response, confirming the APR phenomenon with finer temporal resolution through cross-correlation, generalized additive model smoothing, and pseudotime analyses. Together, static and programmable fixation establish a comprehensive spatiotemporal framework for studying oscillatory translational regulation in single cells.

We then develop graph-based super-resolution protein-protein interaction (GSR-PPI) analysis to predict cancer drug responses from spatial PPI networks in lung adenocarcinoma cells and patient-derived NSCLC tissues, and introduce multiplexed iterative sequential PLA (iseqPLA) to visualize 3D NF-κB supercomplex dynamics. Upstream supercomplexes containing TRAF-5_TRADD and TRAF-5_TRAF-2 undergo rapid dissociation upon cytokine stimulation, inversely correlated with p65 nuclear translocation. Macrophage-fibroblast coculture experiments reveal that CF airway supernatant-exposed macrophages impart a hyperinflammatory phenotype to neighboring fibroblasts through paracrine NF-κB activation. We additionally validate our NF-κB gene panel using scGPT foundation models trained on multi-disease transcriptomic datasets, establishing a mechanistic link between protein-level supercomplex dynamics and transcriptional outputs.

These contributions demonstrate the power of combining spatial multiplexing, temporal reconstruction, and computational modeling to decode immune signaling at single-cell and subcellular resolution, with implications for understanding chronic inflammation in CF and identifying opportune windows for therapeutic intervention.

Date: Friday, May 1st, 2026

Time: 12:00 pm- 2:00 pm Eastern Time

Location: Microsoft Teams

Committee:

• Dr. Ece Erdogmus, Founding Dean, College of Architecture, Art, and Construction, Clemson University (co-advisor and co-committee chair)

• Dr. Eunhwa Yang, School of Building Construction, College of Design, Georgia Institute of Technology (co-advisor and co-committee chair)

• Dr. Jing Wen, School of Building Construction, College of Design, Georgia Institute of Technology

• Dr. Carol Colatrella,  School of Literature, Media, and Communication, Ivan Allen College of Liberal Arts, Georgia Institute of Technology

• Dr. Heidi Diefus-Dux,  Department of Biological Systems Engineering, College of Engineering, University of Nebraska–Lincoln

Title

Enhancing High School Counselors’ Knowledge and Confidence in Promoting Construction Management Careers: A Mixed Method Study

Abstract 

The U.S. construction industry faces a significant and escalating shortage of construction management (CM) professionals driven by an aging workforce, increasing project demand and complexity, and a limited pipeline of students entering CM degree programs. These challenges are further compounded by disconnects between industry workforce needs and the career guidance students receive in high school settings. Although CM offers stable and high-opportunity career pathways, many students remain unaware of the profession. High school counselors, who play a critical role in shaping students’ postsecondary trajectories, often report limited knowledge of CM and low confidence in advising students about CM career pathways. These advising gaps restrict students’ access to CM opportunities and contribute to long-term workforce shortages.

To address these gaps, this study investigates how targeted professional development influences high school counselors’ confidence in advising students about construction management careers. The professional development intervention incorporates structured CM instructional content alongside an experiential learning component using a virtual reality (VR) module derived from the NSF-funded Virtual/Augmented-Reality-Based Discipline Exploration Rotations (VADER-R) project. While VR-based tools have been explored for student recruitment, their application in counselor-focused professional development remains limited.

Grounded in Social Cognitive Career Theory (SCCT), this mixed-methods study focuses on counselor self-efficacy as the primary construct of interest. The study develops and validates a construction management–specific adaptation of the Career Counseling Self-Efficacy Scale (CCSES-CM) to measure counselors’ confidence in advising students about CM pathways. In the quantitative phase, a pre–post design is used to assess changes in counselors’ CM advising self-efficacy following participation in the professional development workshop. In the qualitative phase, semi-structured interviews are conducted with a sample of participants to further explore how the intervention influenced their confidence, perceptions, and advising intentions.

The anticipated contributions of this study are threefold. First, it provides a validated, discipline-specific instrument (CCSES-CM) for measuring counselor self-efficacy in construction management advising. Second, it offers empirical evidence on how targeted professional development can strengthen counselors’ confidence in promoting workforce-critical career pathways. Third, it presents a scalable and adaptable professional development model that can be implemented by school districts, higher education institutions, and workforce development organizations. By strengthening counselors’ capacity to communicate accurate, relevant CM career information, this research supports improved alignment between educational advising systems and the construction industry's workforce needs.

School of Psychology – Ph.D. Dissertation Defense Meeting

Date: Friday, May 1st, 2026

Time: 2:00 PM - 3:30 PM

Location: Na Liu's Dissertation Defense | Meeting-Join | Microsoft Teams

Dissertation Committee Chair/Advisor:
James Roberts, Ph.D. (Georgia Tech)

Dissertation Committee Members:
Audrey Leroux, Ph.D. (Georgia Tech)
Dingjing Shi, Ph.D. (Georgia Tech)
Mark Himmelstein, Ph.D. (Georgia Tech)
Hongli Li, Ph.D. (Georgia State University)

Title: Enhancing Precision in the Generalized Graded Unfolding Model (GGUM) Using Successive Interval Judgements Indicative of Item Location

Abstract: The Generalized Graded Unfolding Model (GGUM) is an ideal-point item response theory model suited to measuring non-cognitive constructs such as attitudes, emotions, and personality. A practical limitation of the GGUM is its requirement for large samples (N > 750) to achieve stable parameter recovery. This dissertation evaluated whether integrating collateral item location information from the Method of Successive Intervals (MSI) into the traditional GGUM could improve parameter recovery and reduce sample size requirements. A Monte Carlo simulation study crossed two sample sizes (N = 300, 750), two response formats (2- and 6-category), and two MSI–trait correlations (r = .50, .80) across 10 replications per cell. Parameter recovery for item discriminations (α), item locations (δ), person locations (θ), and response category thresholds (τ) was evaluated using root mean square deviation (RMSD), posterior standard deviation (PSD), and relative parameter bias (RPB). An empirical comparison was also conducted using attitude-toward-abortion data from 40 items administered to undergraduate participants at the Georgia Institute of Technology. Simulation results showed that estimation quality was driven primarily by sample size and response format, whereas model-related effects were uniformly small and did not favor the new model. In the empirical comparison, posterior summaries for α, θ, and τ were highly similar across models, and δ differences were broadly homogeneous across replications. Overall, adding MSI-based collateral information to the GGUM did not produce consistent or practically meaningful improvements in parameter recovery under the conditions examined.

will propose a doctoral thesis entitled,

Mechanical role of porous mesh for improved dewatering

On

Thursday, April 23 at 10:00 am
EBB Krone 4029

And

Virtually via

https://teams.microsoft.com/meet/24976321543276?p=ANkvYNTOTzqcYcTy16

Meeting ID: 249 763 215 432 76

Passcode: dG9Y6Q9w

Committee
            Dr. Blair Brettmann – School of Materials Science and Engineering (Advisor)
            Dr. Victor Breedveld – School of Chemical and Biological Engineering (Co-advisor)
            Dr. Christopher Luettgen – School of Chemical and Biological Engineering

            Dr. Donggang Yao– School of Materials Science and Engineering

            Dr. Mary Lynn Realff– School of Materials Science and Engineering

Abstract

The papermaking process is highly energy-intensive, with the drying stage accounting for up to 50% of a paper mill’s total energy consumption. To reduce this energy demand, one promising approach is to maximize water removal during the forming and pressing stages prior to evaporation. During pressing, the wet paper is compressed against a felt to drive water out of the paper and into the felt. However, during decompression, the paper web undergoes pore expansion, leading to undesired return of water (rewet). Introducing a stiff, porous mesh layer between the paper and felt improves dewatering by disrupting the liquid bridges that flow back into the fibrous paper. This thesis investigates how the design and intrinsic properties of the mesh porous layer affect dewatering. This includes evaluating the role of mesh geometry, understanding how stress distribution from these porous layers affects the wet fibrous paper, and analyzing transient behavior during compression-decompression cycles.

In the first part of this thesis, I will investigate how the void volume of the mesh provides pathways for fluid to travel between the paper and the felt. I will also study how stacking multiple porous mesh layers between the paper and felt affects dewatering performance. In the second part, I will focus on the role of pressure distribution, specifically how uniform stress and compaction of the paper sheet affect the efficiency of the dewatering process. In the last part, I will examine time as a key process variable and evaluate its influence on the overall pressing system during cyclic pressing. Overall, this research seeks to develop a framework to optimize the mesh–felt system and improve dewatering efficiency by understanding the interactions among layered porous materials during press cycles.

Date: Wednesday, April 29, 2026

Time: 10:30 AM - 12:30 PM, Eastern time (U.S.)

Location: Coda C1115 Druid Hills 

Virtual Meeting (hybrid): https://gatech.zoom.us/j/91333564325

Gennie Mansi

HCC PhD Candidate

School of Interactive Computing, College of Computing 

Georgia Institute of Technology

Committee:

Dr. Mark Riedl (Advisor) - School of Interactive Computing, Georgia Institute of Technology

Mr. Benjamin Sundholm, JD - Tulane Law School, Tulane University

Dr. Naveena Karusala - School of Interactive Computing, Georgia Institute of Technology

Dr. Andrea Parker - School of Interactive Computing, Georgia Institute of Technology

Dr. Agata Rozga - College of Computing, Georgia Institute of Technology

Summary: 

As AI tools are incorporated into high-stakes decision-making environments, such as healthcare and education, people need to act meaningfully in response to AI outputs. This thesis advances actionability—how AI tools and their explanations enable pragmatic action by people in complex sociotechnical environments shaped by power dynamics. I make two central arguments: first, that we can improve people’s ability to act with AI tools by examining how their actions and information needs connect to the ways they care for themselves and others; and second, that understanding how laws and regulations  impact people’s ability to care with AI tools, we can inform the creation and use of AI tools to support people navigating uneven or unknown risks.

This work deepens our understanding of actionability in 5 parts. I discuss how I created a user-centered catalog of information and actions to re-orient the design and evaluation of AI tools around users' needs. Building on this foundation, I draw out the complexities of enabling actionability through an in-depth investigation of physicians' needs, first uncovering how care—including laws, regulations, and collective responsibility for patient well-being—shapes actionability in ways current AI tools’ designs often overlook. Informed by the ways care highlights laws and regulations as a contextual factor, I investigate how doctors' perceptions of potential errors connect to their legal concerns for AI tools. I show that doctors do not connect legal risks to AI tools’ capabilities, and I discuss how this gap may result in unintentional harm in the form of defensive medical practices. To address these gaps, I describe an assets-based, co-design process with lawyers that used visualizations to surface tacit legal knowledge and generate strategies for stakeholders to predict and manage legal risks, revealing how power dynamics shape actionability. Finally, I ground these findings in practice through an analysis of 31 U.S. legal cases, identifying how a complex web of stakeholders and widely deployed AI tools negatively impact patient care, and proposing paths forward through revised liability structures and tools that support legal recourse for patients.

Adrian Choi

Ph.D. Student in Human-Centered Computing (HCC) 

School of Interactive Computing 

Georgia Institute of Technology 

Date: Monday, May 11th, 2026

Time: 10:00 AM-1:00 PM EST 

Location: TSRB 217A or Join via Zoom

Committee 

Dr. Andrea G. Parker (Advisor) – School of Interactive Computing, Georgia Institute of Technology

Dr. Lynn Dombrowski – School of Interactive Computing, Georgia Institute of Technology

Dr. Neha Kumar – School of Interactive Computing, Georgia Institute of Technology

Dr. Brooke Foucault Welles – College of Arts, Media, and Design, Northeastern University

Dr. Christopher Le Dantec – College of Arts, Media, and Design and the Khoury College of Computer Sciences, Northeastern University

Abstract

Free spaces have historically operated as safe learning environments where marginalized youth engage in critical reflection to build civic capacity and mobilize as counter-publics to contest the broader public sphere. However, as youth coalesce around political issues using networked technologies like social media, the formation of these publics becomes actively mediated by commercial technologies. While these technologies allow counter-publics to scale their political discourse, their underlying infrastructure often erodes the conditions necessary for sustaining critical awareness and engaging in constructive deliberation. As such, there is a need to examine how to scaffold technologically-mediated environments that cultivate critically conscious publics in support of an inclusive deliberative democracy.

I present five studies analyzing how sociotechnical assemblages shape critically conscious youth publics. My completed work examines youths' critical consciousness when using social media, investigates how members of a youth empowerment program (YEP) were thwarted even after infrastructuring to sustain a free space, and maps the socio-material conditions necessary for forming healthy networked publics. My proposed work investigates how technology-mediated discourse shapes youths' critical consciousness by analyzing media narratives after police killed Eric Garner. Concurrently, I examine the role of civic data in infrastructuring critically conscious publics within participatory design, aligning public data with YEPs to probe youths' attachments to matters of concern.

This dissertation proposal aims to advance our knowledge of the infrastructuring work involved in scaffolding youths' critical consciousness. By empirically examining YEPs as free spaces alongside the circulation of counter-public discourse online, this work reveals how sociotechnical assemblages can either support or threaten critically conscious publics. Building on these insights, I discuss the precarious nature of infrastructuring these physical and digital spaces, considering the conditions necessary for sociotechnical infrastructures to align with critical pedagogy and support more inclusive democratic deliberation.

Committee:

Dr. Chin-Hui Lee, Advisor

Dr. Moore, Chair

Dr. Anderson

Committee:

Dr. Sathe, Advisor

Dr. Romberg, Chair

Dr. Raychowdhury

Committee:

Dr. AlRegib, Advisor

Dr. Muthukumar, Chair

Dr. Heck

Committee:

Dr. Yu, Advisor

Dr. Hao, Chair

Dr. Mukhopadhyay

Committee:

Dr. AlRegib, Advisor

Dr. Vela, Chair

Dr. Calhoun

Date: Friday, April 24th, 2026  

Time: 2:00 PM – 4:00 PM ET  

Location: Remote (Zoom)  

Virtual Link: https://gatech.zoom.us/j/95008488072  

Bangyuan Liu  

Robotics Ph.D. Candidate  

School of Mechanical Engineering  

Georgia Institute of Technology  

Committee:  

Dr. Frank L. Hammond III (Advisor) – School of Mechanical Engineering, Georgia Institute of Technology  

Dr. Daniel I. Goldman – School of Physics, Georgia Institute of Technology  

Dr. Alper Erturk – School of Mechanical Engineering, Georgia Institute of Technology  

Dr. Layne R. Churchill – Georgia Tech Research Institute  

Dr. Howie Choset – Robotics Institute, Carnegie Mellon University  

Abstract:  

Maneuverability and adaptability are essential capabilities for underwater robots operating in cluttered and constrained environments, where conventional rigid-body vehicles often struggle. This dissertation investigates a family of soft, modular, non-biomorphic swimming robots that leverage mechanical compliance and underactuated design to achieve robust locomotion and interaction with complex environments without reliance on complex sensing or feedback control. We present a series of robotic systems and experiments that progressively explore the relationship between body morphology, actuation, and body–environment interaction. First, we demonstrate how a minimal actuation architecture enables diverse locomotion behaviors, including a novel roll rotation maneuver that provides full three-dimensional maneuverability. Next, we show that intrinsic morphological features, such as compliant, vertebra-like body structures and passive foldable fins, allow the robot to traverse narrow gaps and complex obstacles through passive mechanical adaptation. Building on these insights, we investigate how body scaling and structural parameters influence locomotion performance, revealing the existence of an effective body compliance regime that balances deformation and force transmission for efficient propulsion. Finally, we introduce an actively controllable fin system that extends morphological adaptability through fluidic actuation, enabling dynamic stiffness modulation and improved environmental interaction. Together, this work establishes a design framework for underwater robots that integrates maneuverability, compliance, and adaptability, offering a pathway toward robust operation in complex real-world aquatic environments.

Advisor: Dr. Dimitri Mavris

Milestone: PhD Thesis Proposal

Degree Program: Aerospace Engineering

Title: A Model-Based System-of-Systems Architecting Framework for Lunar Base Master Planning, Evaluation, and Design Space Exploration

Abstract: This research addresses the problem of architecting the lunar base system-of-systems for long-term human presence and driven by stakeholder objectives. As global space agencies, militaries, and commercial companies pursue a sustained presence on the lunar surface, there is a need to establish a physical lunar base capable of supporting scientific, commercial, and national objectives while evolving over time. However, current lunar base site plans lack a comprehensive, systematic process for planning the physical layout and do not account for spatial placement, connectivity, environmental constraints, or how the architecture will grow and adapt under changing requirements and increasing operational complexity. This work approaches the lunar base as a complex system-of-systems and develops a model-based systems architecting framework for generating, evaluating, and evolving lunar base architectures. System architecture decisions are formalized and structured to produce architecture instances through a generative process informed by urban planning practices, capturing the mapping of function to form, as well as spatial/topological and connectivity relationships among the objects of form. The framework leverages model-based systems engineering (MBSE) to capture, query, and trace architectural decisions to the resulting configurations. To enable quantitative evaluation, the framework establishes a connection between MBSE architecture models and multidisciplinary analysis and optimization (MDAO) models through a graph-based transformation mechanism, enabling architecture-specific evaluation of system interactions, resource flows, and infrastructure networks. This allows consistent evaluation of architecture performance while preserving traceability between architectural decisions and analysis results. Given the combinatorial explosion of possible architecture configurations, the design space is characterized by mixed-discrete, hierarchical decision variables and multiple conflicting objectives. To address this system architecture optimization (SAO) problem, population-based multi-objective algorithms augmented with surrogate models are used to efficiently explore the design space and converge to pareto-optimal architectures.  In all, this research establishes a repeatable and systematic methodology for lunar base master planning that integrates architecture generation, transformation for evaluation, and optimization. The methodology enables the analysis and evolution of architecture alternatives for long-term, scalable, and adaptable lunar surface infrastructure.

Date and time: 2026-04-24, 9 AM

Location: Collaborative Visual Environment (CoVE) Room in Weber Space Science and Technology Building

Committee:
Dr. Dimitri Mavris (advisor), School of Aerospace Engineering
Dr. Daniel Schrage, School of Aerospace Engineering
Dr. Perry Yang, School of City and Regional Planning
Dr. Jud Ready, GT Space Research Institute
Dr. Michael Balchanos, School of Aerospace Engineering
, 

Advisor: Dr. Dimitri Mavris

Milestone: PhD Thesis Final Examination (Defense)

Degree Program: Aerospace Engineering

Title: A Design Methodology For Resilient Cislunar Space Domain Awareness Architectures

Abstract: Cislunar space is the next frontier of human space exploration as scientific and commercial missions by national agencies and private entities are driving a renewed race to the Moon. Long-term human presence in this domain depends on supporting infrastructure, of which space domain awareness is a critical component that tackles the detection, tracking, and identification of space objects. SDA has been crucial for near-Earth applications, and the same need will follow as activity in cislunar space grows. Systems resilience as developed for terrestrial infrastructure and for space mission success motivates this work on resilient SDA design. Traditional design processes consider resilience analysis as robustness or disruption analysis after the conceptual design phase, when high-level performance and cost trades are already conducted. Layering resilience onto fixed architectures yields a narrow solution space and weak traceability between resilience outcomes and design decisions. The core technical barrier that blocks resilience from entering the conceptual phase is the modeling complexity of resilience evaluation, where it must be concretely defined, simulated, and measured rapidly enough to participate within the optimization loop. This thesis develops a methodology that brings system-of-systems resilience into the conceptual design of cislunar SDA architectures through a phase-based performance framework, with three research questions derived from the technical gaps. First, a flow-weighted supra-adjacency matrix (FW-SAM) was developed as a graph-based replacement for discrete-event DTN simulation. Encoding sensing, communication, and relay flow in a single sparse matrix, the FW-SAM evaluates downlink efficiency directly from its structure. The second research question addresses graph-based robustness metrics, with sensor performance loss, downlink efficiency loss, and relay efficiency impact derived directly from the FW-SAM. The third research question formulates an analytic resilience integral combining preemption probability, robustness, recovery time, and an adaptation trajectory with a phase-decomposed cost model, where the integral couples resilience investment to the architecture being designed. The validated methods are composed into a multi-objective optimization spanning constellation architecture and resilience phase variables. All-in-One Integrated Resilience Optimization (AIRO) jointly optimizes all variables with, tested against two sequential baselines, with and without resilience optimization. AIRO produced wider Pareto spread on all three objectives and discovered more than twice the architecture classes of the sequential approach, including constellations with lower sensor capability paired with high adaptation investment that the sequential methods cannot find. The resilience objective also drove orbit diversity beyond what performance and cost optimization together could produce. The graph surrogate, the analytic resilience integral, and the architecture-coupled cost model make resilience tractable inside the computational budget of evolutionary optimization, enabling enhanced resilience tradespace exploration for future cislunar SDA systems. 

Date and time: 2026-04-24, 1:30 PM

Location: Code/Cove

Committee:
Dr. Dimitri Mavris (advisor), School of Aerospace Engineering
Dr. Brian Gunter, School of Aerospace Engineering
Dr. Kyriakos Vamvoudakis, School of Aerospace Engineering
Dr. Michael Balchanos, chool of Aerospace Engineering
Dr. Michael Steffens, Draper Laboratory
, 

Advisor: Dr. Dimitri Mavris

Milestone: PhD Thesis Proposal

Degree Program: Aerospace Engineering

Title: A Model-Based Methodology Supporting Commonality Integration in the Lunar Surface Mobility Vehicle Family

Abstract: The goal of the Artemis program is to establish a sustainable lunar settlement, which requires surface mobility vehicles capable of diverse, evolving missions. These use cases support exploration and infrastructure deployment activities. With challenges related to affordability and complexity, NASA has established commonality and interoperability, which relate to sharing common parts and ensuring diverse systems are compatible, as key design principles. Despite the widespread success of leveraging common vehicle platforms in the automotive and aerospace industries, current proposals for lunar surface vehicles remain largely treated as isolated systems rather than a cohesive family of systems where commonality can be leveraged. This fragmented approach overlooks the benefits that commonality can bring, such as reduced development costs and simplified lunar surface logistics and maintenance. However, transitioning to a family of systems perspective introduces complex tradeoffs between commonality and mission-specific performance, necessitating a methodology to identify where commonality should be integrated. To navigate this tradeoff, this research proposes a hybrid approach that leverages the analytical commonality and performance assessments of Product Family Design with the variability management frameworks from Product Line Engineering to support commonality integration decisions. The hybrid approach addresses three main phases of the decision-making process for commonality integration: Modeling of the variability of the lunar surface mobility family of vehicles across its operational, functional, and physical layers of abstraction in a feature model to elicit alternative designs through generative artificial intelligence. Linking simulation instances to the feature model to automatically generate physics-based constraints between these alternatives and ensure the feasibility of design alternatives. Evaluation of the commonality and performance tradeoff and the sensitivity of metric selection for the decision on integrating commonality across parts in different vehicles of the family. Overall, these contributions formulate a methodology that provides decision support on where it is appropriate to leverage commonality.  This research enables physics-driven decisions on a comprehensive description of the alternatives to decide commonality and interoperability. Ultimately, these results support the deployment of robust mobility capability across the lunar terrain through a family of vehicles that addresses the various needs for sustained human habitation on the lunar surface. 

Date and time: 2026-05-01, 9:30 am

Location: COVE (Weber Building)

Committee:
Dr. Dimitri Mavris (advisor), School of Aerospace Engineering
Prof. Daniel Schrage, School of Aerospace Engineering
Prof. Yashwanth Kumar Nakka, School of Aerospace Engineering
Dr. Michael Balchanos, School of Aerospace Engineering
Dr. Hugo Guillermo Chale Gongora, Airbus
, 

will propose a doctoral thesis entitled,

Interface-Engineered MBE of InSe: Early-Stage Growth Control and Hall Sensor Performance

On

Wednesday, April 29 at 11 a.m. (EDT)
Pettit Microelectronics Building Room 102B

and/or

Virtually via MS Teams or Zoom

https://teams.microsoft.com/meet/253152803479756?p=3APd10saNQuRogNWn9

Meeting ID: 253 152 803 479 756

Passcode: Nr2U7Wa9

Committee
            Prof. Eric Vogel – School of Materials Science and Engineering (advisor)
            Prof. Mark Losego– School of Materials Science and Engineering
            Prof. Faisal Alamgir – School of Materials Science and Engineering

            Prof. Alan Doolittle – School of Electrical and Computer Engineering
            Dr. Brent Wagner – Georgia Tech Research Institute

Abstract
               This work aims to enable high-sensitivity, BEOL-compatible InSe Hall sensors by clarifying how the very first layers of film growth set the ultimate limits on mobility and noise. Although indium monoselenide (InSe) offers high intrinsic mobility, a layered van der Waals structure, and a suitable bandgap, MBE-grown films typically underperform exfoliated crystals because defects and competing phases are “baked in” during the first one to three monolayers at the substrate.

               To address this challenge, interface-engineered MBE is used to deliberately shape nucleation and early-stage growth. Metal-precursor layers, chalcogen pretreatments, and time-structured flux schemes are tuned to control nucleation density, phase stability, and grain coalescence, with the goal of producing phase-pure, continuous, low-roughness InSe films within BEOL thermal budgets. Structural and morphological characterization (Raman phase fraction analysis, AFM, TEM) is used to map how these interface-engineering strategies change phase composition, grain size, and interface disorder.

               The influence of substrate bonding and symmetry is further examined by applying these interface-engineered approaches across van der Waals, ionic, and covalent or amorphous platforms. Hall bar devices then relate interface and microstructural metrics directly to Hall mobility, carrier density, sheet resistance, and Hall sensitivity. By combining targeted interface control with quantitative electrical measurements, this thesis seeks to narrow the mobility gap between deposited and exfoliated InSe and to provide actionable guidelines for integrating InSe Hall sensors in future monolithic 3D electronics.

will propose a doctoral thesis entitled,

Multi-scale Computational Investigation of Interfacial Dynamics in Zinc-ion Batteries

On

Tuesday, April 28 at 12:30 p.m.
MRDC 4211

And

Virtually via

https://teams.microsoft.com/meet/24173259120565?p=3RRFE13az92HpDpwO2 

Committee
            Prof. Seung Soon Jang – School of Materials Science and Engineering (advisor)
            Prof. Matthew McDowell – School of Materials Science and Engineering
            Prof. Emma Hu – School of Materials Science and Engineering

      Prof. Marta Hatzell - School of Mechanical Engineering

      Prof. Julia Yang - School of Chemical and Biological Engineering

Abstract

Aqueous zinc-ion batteries (AZIBs) have emerged as a compelling alternative for next-generation energy storage systems due to their inherent safety, cost-effectiveness, and environmental sustainability. However, the practical deployment of AZIBs is severely hindered by fundamental challenges at the zinc anode interface, including uncontrolled dendritic growth, hydrogen evolution reactions (HER), and surface corrosion. This research proposes a multi-scale computational framework—integrating density functional theory (DFT), molecular dynamics (MD), and machine-learned interatomic potentials (MLIP)—to design high-performance zinc anode protective layers and optimize electrolyte systems.

The first part of this work investigates the atomic-scale mechanisms of a zinc-doped polypyrrole (Zn-doped PPy) conductive polymer as a functional protective layer. Using DFT calculations, we analyze the structural and electronic properties of PPy, demonstrating that Zn doping at nitrogen sites significantly enhances electrical conductivity by reducing the bandgap and creates potent nucleation sites for uniform Zn deposition. Furthermore, the synergistic effect of incorporating carbon nanotubes (CNTs) into the Zn-doped PPy matrix is explored, focusing on the enhanced mechanical adhesion and the formation of a stable, conductive network that effectively regulates the electric double layer.

The second phase of the research focuses on electrolyte engineering through MD simulations. We examine the solvation and desolvation dynamics of zinc ions in aqueous environments, specifically investigating the impact of organic additives such as ethylene glycol (EG). By analyzing the competitive coordination between water molecules and additives within the Zn2+ solvation shell, we evaluate the advantages and trade-offs of additive engineering in suppressing water-induced side reactions while maintaining high ionic conductivity.

Finally, to bridge the gap between atomic-level interactions and mesoscopic electrochemical behavior, MLIPs are developed and deployed to simulate large-scale zinc deposition processes. This integrated approach allows for a comprehensive analysis of dendrite inhibition and the evolution of the electrode-electrolyte interface. The findings of this research will provide fundamental theoretical insights and design principles for developing dendrite-free, long-life aqueous zinc-ion batteries for large-scale energy storage applications.

Advisor: Dr. Sarah Li

Milestone: MS Thesis Proposal

Degree Program: Aerospace Engineering

Title: Distributed MPC for momentum counter-balanced and zero-impulse contact of a free-spinning satellite

Abstract: In on-orbit robotics, a servicer satellite's ability to make contact with a free-spinning target satellite is essential to completing most on-orbit servicing (OOS) tasks. A tumbling target satellite with angular momentum components across multiple body axes may require multiple servicer satellites contacting in different planes to achieve full detumble. This manuscript develops a cooperative distributed model predictive control (DMPC) framework that coordinates the maneuvers between multiple servicer satellite agents to achieve zero-impulse contact with a free-spinning target satellite. Each servicer internally runs a hierarchical MPC controller that coordinates between two separately actuated modules of the servicer satellite: (1) a moment generation module and (2) a manipulation module. The distributed outer layer coordinates across servicers by enforcing consensus on the shared target state. We analyze the computational complexity of the large-scale centralized optimization problem and the smaller-scale distributed subproblems independently and then analyze how the number of agents scales with computational complexity. We evaluate the performance of the DMPC framework by simulating multi-servicer zero-impulse contact scenarios with a free-spinning target satellite via numerical Monte Carlo (MC) trials and comparing the simulation results with the centralized MPC framework. We validate the multi-servicer contact scenario using MuJoCo to verify that the coordinated detumble maneuver satisfies zero-impulse contact constraints across all servicer-target contact points.

Date and time: 2026-04-23, 11:00 AM

Location: Montgomery Knight 317

Committee:
Dr. Sarah Li (advisor), School of Aerospace Engineering
Dr. Sarah Li, School of Aerospace Engineering
Dr. Glen Chou, College of Computing and School of Aerospace Engineering
Dr. Panagiotis Tsiotras, School of Aerospace Engineering
Dr. Koki Ho, School of Aerospace Engineering
NA, NA

Advisor: Dr. John Dec

Milestone: MS Thesis Proposal

Degree Program: Aerospace Engineering

Title: Improving Atmospheric Entry Environment Prediction Using Machine-Learning Surrogate Models

Abstract: The expansion of the space economy, driven by both private and public sectors, has increased the demand for atmospheric entry prediction. Accurate predictions in early mission analysis enable guidance and control design, establishment of mission requirements, landing footprint estimation, and space debris risk assessment. Current early mission analysis tools face a fundamental trade-off: simplified models are computationally fast but neglect key physics such as aerodynamics and ablation; high-fidelity multi-physics simulations are accurate but computationally expensive for uncertainty studies. This work proposes a mission-driven framework that bridges this gap by coupling a 3-DOF non-planar trajectory propagator with FUN3D computational fluid dynamics and FEAR ablation analysis to generate high-fidelity training data, which is then used to develop machine-learned surrogate models for aerodynamic coefficients and ablation response. Training data is concentrated near a nominal trajectory; its plausible dispersions with altitude, velocity, and vehicle geometry are used as inputs, and lift coefficient, drag coefficient, mass loss, and frontal area recession are used as outputs. The coupled Trajectory-CFD Solver has been developed and validated against the Stardust sample return capsule re-entry. The framework predicts the landing site within 41 miles of the actual location, reducing the prediction error by nearly half compared to a conventional low-fidelity trajectory propagator. Remaining work includes data generation improvements, surrogate model training and deployment, integration of the surrogates into the trajectory propagator, and Monte Carlo uncertainty quantification of landing footprints. The resulting framework will demonstrate how mission-specific surrogate modeling can preserve the fidelity of high-fidelity solvers at a fraction of the computational cost, advancing early-phase reentry mission analysis.

Date and time: 2026-05-04, 11:00 AM

Location: ESM 108

Committee:
Dr. John Dec (advisor), School of Aerospace Engineering
Dr. Krish Ahuja, School of Aerospace Engineering
Dr. Alvaro Romero-Calvo, School of Aerospace Engineering
, 

will defend a master's thesis entitled,

Aluminum Isopropoxide: A Versatile Crosslinker for the Development of Organic-Inorganic Hybrid Materials 

On

Tuesday, April 21 at 3:00 p.m.
EBB Conference Room 5029

and/or

 Virtually via MS Teams or Zoom

https://teams.microsoft.com/l/meetup-join/19%3ameeting_ZWI4NmZhNjItN2YxYi00N2UwLWJhMjYtNDM4NGRiMDRhMDg4%40thread.v2/0?context=%7b%22Tid%22%3a%22482198bb-ae7b-4b25-8b7a-6d7f32faa083%22%2c%22Oid%22%3a%22d94c616a-028c-45a9-9c4a-6d0ca86cc35e%22%7d

Committee
            Prof. Blair Brettmann – School of Materials Science and Engineering, School of Chemical and Biomolecular Engineering (Advisor)
            Prof. Lukas Graber – School of Electrical and Computer Engineering
            Prof. Juan-Pablo Correa-Baena – School of Materials Science and Engineering

Abstract

Organic-inorganic (OI) hybrid materials are a class of materials combining organic and inorganic components at the molecular level, creating a synergistic effect in properties that is greater than the sum of each part. Epoxy resins are known for having high chemical and corrosion resistance, high impact resistance, toughness, and outstanding adhesion. They are widely used as adhesives, high-performance composite matrices for aerospace and sporting goods, and protective coatings against corrosive and radiative environments. By incorporating inorganic materials into the epoxy matrix at the molecular level, properties of the organic epoxy and the inorganic fillers can be covalently connected, decreasing the risks of phase separation and leading to a homogenous hybrid material. Existing epoxy OI hybrid materials typically rely on silicon as the inorganic component and require the addition of hardeners, catalysts, or solvents to generate a crosslink network. Aluminum isopropoxide (AIP) presents an opportunity for the development of a two-part epoxy – AIP hybrid system. AIP takes on a dual-functional role as both an initiator for the ring-opening polymerization (ROP) of the epoxide ring and as an inorganic crosslinking center.

In this work, I demonstrate that AIP concentration influences the thermal properties of the epoxy – AIP network and can match or exceed that of epoxy resins cured with traditional hardeners. Curing was carried out under high pressure (300 psi) to suppress bubbling from the volatilization of free isopropanol generated during the reaction, enabling the formation of consistent, defect-free hybrid materials.

This work analyzes and compares the results of three epoxy systems: a commercial epoxy resin (EC 1159A), 2,2-Bis(4-glycidyloxyphenyl)propane / Bisphenol A diglycidyl ether (DGEBA), and Bis(7-oxabicyclo[4.1.0]heptan-3-ylmethyl) adipate  (BECHMA). Thermal properties of the commercial epoxy – AIP, DGEBA – AIP, and BECHMA – AIP systems cured at varying AIP concentrations revealed the effects AIP has on the crosslinking network and the thermal stability of the systems. At 15 wt% AIP, a more homogenous crosslinked network formed, evidenced by the higher reported values for glass transition and onset of thermal degradation, while increased AIP concentrations resulted in hybrids with a greater resistance to total thermal degradation and higher thermal stability.

Overall, this work highlights the strong initiation properties of AIP towards the ROP reaction of epoxide rings. Combined with its tri-functional nature, AIP is a powerful cross-linker that is applicable across the investigated systems and can be used to cure epoxy resins with relative simplicity, resulting in consistent effects on the thermal properties.

Advisor: Dr. Timothy Lieuwen

Milestone: PhD Thesis Proposal

Degree Program: Aerospace Engineering

Title: Characterization and Modeling of Nonlinearly Interacting Thermoacoustic Modes in a High Pressure Gas Turbine Combustor

Abstract: Thermoacoustic instability is a widely-recognized performance limitation found in a variety of combustion systems, and pushing operations towards higher power and reduced emissions has created conditions where these instabilities are more prevalent. Because of their importance in determining operability limits and structural loads, the effort to understand, model, and predict these pressure and heat release oscillations has garnered much attention. Decades of experimental analysis and modeling work has resulted in a very robust understanding of the linear behavior and stability of thermoacoustic modes. However, numerous experiments have noted the presence of limit cycles, intermittency, and chaos which cannot be understood through linear analyses. Further complicating these systems is the presence of multiple linearly-unstable modes whose coupling can cause yet more complex behaviors. This work proposes to contribute to the understanding of the nonlinear dynamics of coupled thermoacoustic modes through both experimental characterization and model building. A novel experimental facility will explore both the steady-state and transient behavior of a system with multiple interacting modes. By using a hydraulically-actuated variable-length exhaust, this test setup can precisely control acoustic mode shapes without altering operating conditions. A data campaign will characterize the dynamics of global and spatially-resolved heat release and acoustic pressure in this rig at high speed across gas-turbine relevant operating conditions. To extract more generalizable information about the underlying physics in the system, a reduced order model will be constructed to reproduce the observed behaviors. By connecting model terms to physical phenomena, the model will help uncover the mechanisms underpinning the system dynamics. In turn, this model and the associated learnings will aid in the modeling of thermoacoustics in pursuit of better design and more predictable operation of gas turbine combustors.

Date and time: 2026-04-28, 9:30 am - 11:30 am

Location: Food Processing Technology Building - Auditorium 102

Committee:
Dr. Timothy Lieuwen (advisor), School of Aerospace Engineering
Dr. Timothy Lieuwen, School of Aerospace Engineering
Dr. Benjamin Emerson, School of Aerospace Engineering
Dr. Keegan Moore, School of Aerospace Engineering
, 
, 

Student Name: Yidong Hua

Thesis Title: Assessing the feasibility and impacts of full electrification on the pulp and paper industry

Thesis Advisor: Marta Hatzell

Thesis Co-Advisor: Micah Ziegler

Committee Members: Sankar Nair (ChBE), Mattew Realff (ChBE), Chris Luettgen (ChBE), Joe Bozeman (Civil and Public Policy)

Date: 04/28/2026

Time: 3:00 - 5:00 PM

Location: MRDC 3515

(Advisor: Prof. Seung Soon Jang & Prof. W. Hong Yeo)

will propose a doctoral thesis entitled,

Development of soft materials and kirigami-based mechanical decoupling for motion-artifact reduction in wearable electronics

On

Tuesday, April 28th at 10:00 am

Marcus Nanotechnology Research Center Room #2107 

Abstract

Wearable bioelectronics enable continuous physiological monitoring but are fundamentally limited by motion artifacts arising from mechanical instability at the electrode–skin interface. This issue is particularly severe in dynamic conditions and non-conformal wearable systems, where motion-induced noise overlaps with low-frequency bio-signals such as EMG and EOG, making conventional structural or computational approaches insufficient.

In this work, we present a multi-scale strategy for motion artifact suppression that integrates structural design, geometrical optimization, and material-level damping. Kirigami-patterned structures are first introduced to achieve strain isolation by redistributing mechanical deformation away from the sensing interface, enabling stable EMG acquisition in freely moving ALS animal models. These structures are further optimized through systematic analysis of kirigami geometries, where hierarchical fractal patterns demonstrate improved strain isolation, reduced mechanical instability, and enhanced signal quality in dynamic applications.

To address non-conformal wearable systems, we develop a material-based approach using polyborosiloxane (PBS), a viscoelastic polymer with dynamic boroxine networks. Molecular dynamics simulations and rheological experiments reveal that boroxine-rich networks provide frequency-stable energy dissipation, enabling effective suppression of motion artifacts in EOG systems.

This work establishes a comprehensive framework that combines structural and material design strategies to enable robust bio-signal acquisition across diverse wearable platforms.

Committee

• Prof. Antonio Facchetti – School of Materials Science and Engineering
• Prof. Shucong Li – School of Materials Science and Engineering
• Prof. Peter Hesketh – School of Mechanical Engineering

College of Design – School of Building Construction – PhD Proposal Defense – Steven Kangisser

Title: Abstraction, Presence, and Transfer; A Framework for Visualization Design in Digital Demonstrators for Equipment-Intensive STEM Education

Dissertation advisory committee:

· Dr. Javier Irizarry, School of Building Construction, Georgia Institute of Technology

· Dr. Eunhwa Yang, School of Building Construction, Georgia Institute of Technology

· Senior Lecturer Tim Purdy, School of Industrial Design, Georgia Institute of Technology

· Dr. Maribeth Gandy Coleman, Institute of People and Technology, Georgia Institute of Technology

· Dr. Jamie Gorman, Human Systems Engineering, Arizona State University

Date and Time: Wednesday, April 22, 2:30-4:00 PM EST

Location: Cadell Building, Conference Room, Floor 2, 280 Ferst Dr., Georgia Institute of Technology, Atlanta, GA 30030

Virtual: https://teams.microsoft.com/meet/221081871285606?p=OCp7eQsZV1ozOHW5la

Abstract: Construction management programs face a persistent instructional challenge: students must develop competency with complex, expensive instruments that most programs cannot procure. Digital demonstrators have been proposed as viable alternatives, yet the field lacks empirical guidance on how to design the visualizations embedded within them. This dissertation addresses that gap using laser scanning instruction as its primary case. The dissertation is designed to contribute an empirical comparison of distinct visualization approaches in an instrument-level digital demonstrator using a behavioral transfer measure.

The study is organized around three research questions corresponding to three identified gaps in literature. RQ1 examines how visualization design within a screen-based digital demonstrator affects the risk of visualization dependency. RQ2 investigates how the extent of demonstrator exposure influences physical scanner skill transfer, as measured by setup time. RQ3 asks what instructional outcomes a full-scale immersive VR demonstrator produces relative to the screen-based tool, and whether VR presence supports deeper schema formation or amplifies scaffold dependency.

Two theoretical frameworks govern the study. Cognitive Load Theory (CLT) provides the primary lens, framing dependency risk as a consequence of over-scaffolding that bypasses the germane processing required for durable schema formation. The Cognitive Affective Model of Immersive Learning (CAMIL) governs the VR strand, positioning presence and agency as mediators of immersive learning benefits while acknowledging that immersive novelty may also increase extraneous load.

The study employs mixed methods, pragmatist design across seven sequential phases. A screen-based demonstrator was developed incorporating two visualization conditions. The first being a lower-fidelity abstract circle grid that required active conceptual inference. The second is a higher-fidelity apple point cloud designed to make the resolution-quality relationship perceptually explicit. Physical scanner setup time served as the primary behavioral outcome, chosen for its independence from self-report and its sensitivity to dependency effects.

RQ2 is supported by the strongest evidence. Graduate teams with full demonstrator access completed physical scanner setup much more quickly as compared to a more limited-access cohort. This dose-response pattern was independently replicated in an undergraduate cohort at Georgia Tech, substantially strengthening the evidentiary basis. RQ1 is supported by preliminary evidence in which all five teams trained on the higher-fidelity visualization exhibited dependency when transitioning to the physical scanner. That finding requires replication before it can be treated as confirmed. RQ3 remains unresolved pending the next phase of this research which will introduce a VR visualization, testing emersion and presence as a level of visualization. This VR visualization will intentionally have a lower level of photorealism as compared to either of the 2-dimensional prototypes.

Completed phases have yielded five confirmed or provisionally confirmed screen-based design criteria and three provisional VR criteria, constituting an empirically grounded framework for visualization design in digital demonstrators for instrument-intensive STEM education.

Committee:

Dr. Ayazi, Advisor

Dr. Shaolan Li, Chair

Dr. Ansari

Committee:

Dr. Shaolan Li, Advisor

Dr. Gu, Chair

Dr. Sathe

Committee: 

Dr. Peterson, ECE, Advisor   

Dr. Naeemi, ECE

Dr. Losego, MSE

Date: Thursday, April 23, 2026

Time: 10:00 AM – 12:00 PM (Eastern Time)

Location (Virtual): https://gatech.zoom.us/j/95898773125

Mingyu Guan

Ph.D. Student

School of Computer Science

Georgia Institute of Technology

Committee Members

Dr. Taesoo Kim (Advisor), School of Cybersecurity and Privacy, Georgia Institute of Technology

Dr. Anand Iyer (Co-advisor), School of Computer Science, Georgia Institute of Technology

Dr. Ada Gavrilovska, School of Computer Science, Georgia Institute of Technology

Dr. Kexin Rong, School of Computer Science, Georgia Institute of Technology

Dr. Jay Stokes, Microsoft Research

Abstract

Graph neural networks (GNNs) are increasingly deployed in high-stakes domains such as fraud detection, traffic prediction, and social network analysis. However, real-world graphs are heterogeneous, dynamic, and privacy-critical, posing challenges in structural complexity, scalability, and verifiability across the model lifecycle. This thesis argues that graph learning problems contain inherent structural regularities that, when explicitly leveraged in model and system design, enable scalable and verifiable graph learning across critical stages of the model lifecycle. First, we present HetTree, a scalable heterogeneous GNN that exploits the natural tree hierarchy among metapaths by constructing a semantic tree representation and introducing a subtree attention mechanism to capture hierarchical relationships with low computation and memory overhead. Second, we present ReD, a system for scalable dynamic GNN training that leverages independence across snapshot sequences to enable sequence-parallel training without cross-machine communication, supported by sequence-first mini-batching and a two-level cache store. Third, we present TAITEE, a system that establishes verifiable training provenance by integrating training recording with Confidential Computing, recording comprehensive provenance via standard training APIs and generating cryptographically signed training certificates from Trusted Execution Environments with minimal overhead. Together, the three contributions span the critical graph learning lifecycle: what to compute, how to compute it efficiently, and whether the training was conducted as claimed — providing scalable and verifiable foundations for deploying graph learning models in practice.

Master of Science in Biology

in the

School of Biological Sciences

Tianyi Ye

Will defend his thesis

“Functional Roles of Rad52 and RPA in Inverse RNA Strand Exchange”

22, April, 2026

10: 00 AM (EST) in EBB-1-5024.

https://gatech.zoom.us/j/92351872480?pwd=5Yq8gGcTDl5HY45siEZe6qIpwpUj8w.1

Passcode: [To be shared on the day of]

Thesis Advisor:

Dr. Francesca Storici

School of Biological Sciences

Georgia Institute of Technology

Committee Members:

Dr. Yury O. Chernoff

School of Biological Sciences

Georgia Institute of Technology

 Dr. Alexander Mazin

Department of Biochemistry and Structural Biology

UT Health San Antonio

Abstract: RNA-templated DNA repair has emerged as an alternative pathway for maintaining genome stability following DNA double-strand breaks (DSBs). Rad52 is a key factor in homologous recombination (HR), where it promotes strand annealing and coordinates repair intermediates. Previous in vitro studies have shown that interaction between Rad52 and replication protein A (RPA) influences inverse RNA strand exchange, a reaction in which RNA can guide DNA repair. These findings suggest a role for Rad52–RPA interaction in RNA-templated DNA repair; however, whether this mechanism operates in vivo remains unclear.

In this study, we developed a yeast-based system to investigate the role of Rad52–RPA interaction in RNA-templated DNA repair. Using a chromosomal reporter assay in which restoration of gene function serves as a quantitative readout, we measured repair frequency as a proxy for inverse RNA strand exchange and evaluated Rad52 variants with disrupted RPA interaction. This approach enables direct assessment of the requirement for Rad52–RPA interaction in RNA-dependent repair in vivo and provides new insight into RNA-mediated DNA repair mechanisms.

Doctor of Philosophy in Quantitative Biosciences

in the

School of Biological Sciences

Christopher Zhang

Will defend his thesis

“THE COSTS AND CONSTRAINTS OF BIOLOGICAL COMPLEXITY”

April 22nd, 2026

2:00 PM

DALNEY 180

https://gatech.zoom.us/j/98792482360?pwd=DLP1U5pLjGt6ssC3iJWXqxL24ylm5f.1

Meeting ID: 987 9248 2360

Passcode: Crisis

Thesis Advisors:

Dr. Brian K. Hammer

School of Biological Sciences

Georgia Institute of Technology

Dr. William C. Ratcliff

School of Biological Sciences

Georgia Institute of Technology

Committee Members:

Dr. Samuel P. Brown

School of Biological Sciences

Georgia Institute of Technology

 Dr. Peter J. Yunker

School of Physics

Georgia Institute of Technology

 Dr. Lauren Speare

School of Biological Sciences

Georgia Institute of Technology

Abstract:

The costs and constraints of biological complexity are central to understanding how sophisticated traits and communities evolve and persist. In this dissertation defense, I will present two stories involving the Type Six Secretion System (T6SS), a complex bacterial apparatus that allows a bacterium to kill an adjacent competitor with remarkable efficiency. T6SSs have been shown to strongly impact microbial population dynamics, and there is an ongoing effort to understand exactly how this occurs. 

First, I will describe my efforts to precisely measure the metabolic cost of utilizing the T6SS, showing that this cost is significantly lower than prior expectations. Then, I will present a combined effort to scan the entire bacterial phylogenetic tree to identify which and how many bacteria actually possess the T6SS, revising a widely cited estimate that has stood for nearly two decades. Together, these two studies refine our understanding of the T6SS as a whole and reshape how we think about the role of bacterial weaponry in shaping microbial dynamics over evolutionary timescales.

Mengqi Lou

ACO PhD student

School of Industrial and Systems Engineering

Date: April 22, 2026
Time: 12:00 PM – 2:00 PM (EST)
Location: Groseclose 226, Georgia Tech Campus

Zoom: TBA

Thesis: https://drive.google.com/file/d/1dUQuAif01CY-8o2ccTlNoDsVtQHR-Pqx/view?usp=sharing

Committee:

Dr. Ashwin Pananjady (Advisor), Schools of Industrial and Systems Engineering & Electrical and Computer Engineering, Georgia Tech
Dr. Cheng Mao (Reader), School of Mathematics, Georgia Tech
Dr. Will Perkins, School of Computer Science, Georgia Tech
Dr. Justin Romberg, School of Electrical and Computer Engineering, Georgia Tech
Dr. Guy Bresler, Department of Electrical Engineering and Computer Science , MIT

Abstract:

The task of learning the underlying parameters of a statistical model from noisy samples is ubiquitous in modern signal processing and data science. Both computational and statistical challenges arise, especially in high-dimensional settings where the number of parameters is comparable to (or exceeds) the sample size. On the computational side, iterative algorithms are commonly used to fit complex models to random data, but their design and analysis are often guided by worst-case upper bounds that may not reflect practical performance. On the statistical side, classical information-theoretic limits on sample complexity or signal-to-noise ratio may be unattainable by any polynomial-time procedure, making these limits an impractical benchmark for modern high-dimensional problems.

In this thesis, we discuss two general frameworks that address these computational and statistical challenges. In the first part, I will present a toolkit that yields sharp, iterate-by-iterate characterizations of solution quality for complex iterative algorithms on several non-convex model-fitting problems with random data. In the second part, I will present a toolkit to derive average-case “reductions’’ between different statistical models, illustrating how such reductions reveal the computational limits of several structured high-dimensional problems.

Student Name: Jeong Soo (Jennifer) Lee

Thesis Title: Developing Multivalent Nanoparticle Vaccines and Antibody-based Therapeutics

Thesis Advisor: Ravi Kane

Thesis Co-Advisor: N/A

Committee Members: John Blazeck (ChBE), Mark Prausnitz (ChBE), Corey Wilson (ChBE), Ronghu Wu (School of Chemistry and Biochemistry)

Date: 04/23/2026

Time: 2 PM

Location: EBB Children's Healthcare of Atlanta (CHOA) Seminar Room

Advisor: Dr. Dimitri Mavris

Milestone: PhD Thesis Proposal

Degree Program: Aerospace Engineering

Title: Intent-Conditioned Trade-Space Exploration and Decision Support for Early-Stage Aerospace System Design

Abstract: Early-stage aerospace design is the period during which fundamental architectural choices are made under the greatest uncertainty. The decisions made in this phase, covering performance budgets, structural concepts, and propulsion configurations, are rarely substantially revised thereafter, yet the processes that govern them provide no structured means of revisiting them coherently as requirements, assumptions, and knowledge continue to evolve. A review of the literature reveals that, despite mature processes for feasibility assessment and requirements management, early-stage aerospace design lacks a systematic means of exploring, justifying, and revisiting design decisions as the knowledge that governs them continues to evolve. To address these, this thesis proposes a methodology centered on the concept of "system intent", the formally encoded combination of prioritized requirements, decision rationale, explicit assumptions, and tacit knowledge, and its systematic connection to multidisciplinary trade-space exploration and decision justification. The methodology is developed across three successive research areas and evaluated through a multi-domain, commercial-aircraft conceptual design demonstration that involves sequential changes to requirements and assumptions. The result is a pipeline that equips designers to formally encode the reasoning behind design decisions in a way that actively drives trade-study formulation, to respond to requirement and assumption changes through targeted updates rather than complete reconstruction, and to assess, at any point in the conceptual phase, which assumptions most threaten the stability of the current design selection and what would need to change for an alternative decision to be warranted, capabilities that current practice does not support. This work provides a structured foundation for design decision-making in complex aerospace programs where requirements, priorities, and knowledge are subject to continuous revision.

Date and time: 2026-04-23, 8AM - 11AM

Location: Collaborative Design Environment (CoVE), Weber (SST II)

Committee:
Dr. Dimitri Mavris (advisor), School of Aerospace Engineering
Prof. Daniel Schrage, School of Aerospace Engineering
Prof. Kai James (Pending), School of Aerospace Engineering
Dr. Elena Garcia, School of Aerospace Engineering
Dr. Jorge Camacho-Casero, Airbus
, 

Date:  Monday, 27th April

Time: 10.30am (EST)

Zoom link: https://gatech.zoom.us/j/5430145654?pwd=E6D8k4UWjLaOURjO0RyzvsyMVFtyOx.1&omn=95297390163

Committee members:

Dr. Christopher Heil (advisor)

Dr. Michael Lacey

Dr. Doron Lubinsky

Dr. Michael Damron

Dr. Kasso Okoudjou

Abstract:

The first chapter focuses on density bounds. Beurling density bounds for basis-type systems have been studied for several decades, in the frameworks of exponential systems, Gabor systems, wavelets and more general reproducing kernel Hilbert spaces. In the work of Olevskii and Ulanovskii (2009), and Nitzan (2024), necessary density conditions were established for a new class of systems known as ‘near-uniformly minimal’ exponentials. Near-uniform minimality is a generalization of uniform minimality (which itself can be thought of as an analogue of linear independence). We deduce analogous necessary density conditions for near-uniformly minimal Gabor systems. Completeness is a dual property to minimality. Much in the same way, Nitzan (2024) introduced a dual notion of ‘uniformly complete’ exponentials. She also proved necessary density conditions in this case, using a duality argument, connecting it to near-uniform minimality. We establish analogous density bounds for uniformly complete Gabor systems. However, we use a different technique, since duality fails for Gabor systems over the real line. We extend many of our results relating to near-uniform minimality/completeness, to select reproducing kernel Hilbert spaces.

In the next chapter, we study woven weighted exponential systems. A classical problem that has been extensively studied, is the basis structure problem of weighted exponentials. We consider analogous questions in the woven setting. As the name suggests, given two weights f and g, a weaving is a sequence where some of the vectors are taken from the first weighted exponential system, and the remaining are taken from the other. In our setup the standard ordering is imposed on the integers. Casazza, Cabrelli and Molter have studied woven systems in abstract Hilbert space. Our focus is specifically on woven weighted exponentials in L^2([0,a]). We obtain a complete characterization of wovenly complete systems, based on the properties of the two weights f and g. We also find a characterization for woven Bessel systems and orthonormal bases. In the case of woven frames/Riesz bases, we obtain sufficient conditions, involving a suitable perturbation requirement on g relative to f.

In the final chapter, we study oblique dual frames and weighted frames in Hilbert space. Given two subspaces of a Hilbert space, W and V, such that W and the orthogonal complement of V form a direct sum, we can define oblique duals. Oblique duals generalize frame duals, in the sense, that for a frame {f_n} in a subspace W, an oblique dual is a frame {g_n} living in V , satisfying a decomposition condition for elements in W.  We define a more general notion known as ‘weighted oblique dual frames’, where the usual oblique dual decomposition becomes a weighted sum. When the weight sequence is {1}, one gets the usual oblique dual. Analogous to the known results for oblique duals, we obtain a complete characterization of weighted oblique duals. We also study their existence when the weight sequence may be degenerate. Another avenue we explore, includes proving weighted potential and coherence results for weighted oblique dual frames. Finally, we establish a connection between the existence of weighted oblique dual frames, and the weighted frame problem. 

Adam Coscia

Ph.D. Candidate in Human-Centered Computing 

School of Interactive Computing, College of Computing

Georgia Institute of Technology 

https://adamcoscia.com/

Date: Monday, April 27, 2026

Time: 10AM - 12PM Eastern time (U.S.)

Location: TSRB room 334 (VIS Lab) – just walk in, show your BuzzCard to the concierge if asked

Virtual Meeting (hybrid): https://gatech.zoom.us/j/3100254613?pwd=QWlKajNkOWlPbWkxR3N5MkZsTE9FZz09 

Committee

Dr. Alex Endert - School of Interactive Computing, Georgia Institute of Technology

Dr. Duen Horng (Polo) Chau - School of Computational Science & Engineering, Georgia Institute of Technology

Dr. Cindy Xiong Bearfield - School of Interactive Computing, Georgia Institute of Technology

Dr. Yalong Yang - School of Interactive Computing, Georgia Institute of Technology

Dr. Scott Crossley - Department of Special Education, Vanderbilt University

Abstract

The advent of powerful new large language models (LLMs) has catalyzed a surge in LLM-powered educational technologies, enabling transformational advances that can empower learner agency, deliver personalized study materials, and promote active learning. Yet persistent pedagogical risks, from bias and hallucinations to unfair grading and misalignment with instructional goals, highlight a critical technology gap. Existing tools for selecting, fine-tuning, and evaluating LLMs are not designed to address the unique challenges of educational contexts, making it difficult for data scientists to detect and mitigate the potential pedagogical risks of prematurely deploying LLM-powered educational technology.

This thesis addresses this gap by introducing human-in-the-loop visual analytics approaches that integrate automated analysis with interactive visualizations, enabling data scientists to more effectively discover, understand, and address pedagogical risks throughout the LLM development lifecycle. We organize these contributions under four main thrusts:

(1) Uncovering Harmful Biases and Stereotypes in LLM Selection: We introduce KnowledgeVIS, a visual analytics system that enables interactive exploration of fill-in-the-blank prompts to surface latent biases, stereotypes, and learned associations in foundation LLMs. By supporting comparative analysis across models, KnowledgeVIS empowers data scientists to make more informed model selection decisions prior to deployment, revealing risks that are often obscured by traditional benchmark-driven evaluation.

(2) Diagnosing LLM Decision-Making During Fine-Tuning: We present iScore, a human-in-the-loop visual analytics system for interpreting how LLMs make scoring decisions in educational tasks such as automatic writing assessment. By linking internal model representations with input perturbations and output variations, iScore enables data scientists to diagnose unintended decision-making criteria, uncover model sensitivities, and iteratively refine fine-tuning strategies to better align with pedagogical objectives.

(3) Measuring and Visualizing LLM Trustworthiness in Evaluation: We propose a novel framework for operationalizing LLM trustworthiness as a set of interpretable, pedagogically grounded metrics, coupled with visualizations that make these risks traceable within model outputs. Through a co-designed evaluation workflow, we demonstrate how these metrics improve the consistency, transparency, and defensibility of expert decision-making, while surfacing complex trade-offs that cannot be captured by traditional performance measures alone.

(4) Broadening Access to Visual Analytics in Education: We contribute TrustyVis, an open-source Python library that encapsulates the visual analytics techniques developed in this thesis into modular, reusable components. By lowering the barrier to integrating interactive visualizations into existing machine learning workflows, TrustyVis enables scalable and practical adoption of human-in-the-loop approaches for evaluating and improving LLM-powered educational systems.

Through a multi-year longitudinal co-design process with data scientists as well as several deployments and integrations into real-world educational settings, this thesis demonstrates how human-in-the-loop visual analytics can transform opaque LLM pipelines into transparent, iterative, and trustworthy development processes, ultimately supporting the responsible integration of LLMs into high-stakes learning environments. The outcomes of this thesis have been disseminated through multiple publications in top journals and conferences, advancing the state of the art in visual analytics, human-computer interaction, artificial intelligence, and educational technology by establishing new methods, systems, and design principles for making LLM behavior interpretable in pedagogical contexts.

Committee:

Dr. Shimeng Yu, ECE, Chair, Advisor

Dr. Muhannad Bakir, ECE

Dr. Saibal Mukhopadhyay, ECE

Dr. Tushar Krishna, ECE

Dr. Yingyang (Celine) Lin, CoC

Committee: 

Dr. Raychowdhury, Advisor     

Dr. Li, Chair

Dr. Sathe

Committee: Dr. Carl DiSalvo (Chair), Dr. Richmond Wong, Dr. Noura Howell, Dr. Christopher Le Dantec & Dr. Sarah Fox

Summary: In the context of the environmentally devastating and socially exploitative global fashion industry, online secondhand economies in the United States and other Global North contexts are often celebrated as paths toward sustainable consumption, ethical recirculation, and entrepreneurial opportunity. This dissertation argues that online resale is better understood as a form of platform-mediated labor embedded within imperial surplus economies and colonial racial capitalism. It focuses on the imperial middle, between exploited labor at the point of production and the colonial geographies of discard, where resellers source, classify, aestheticize, and recirculate surplus goods through platforms such as Poshmark, eBay, and Depop.

Drawing on five years of mixed-methods ethnographic engagement, participatory arts-based research, and autohistoria-teoría, I examine how these conditions are lived and negotiated. I develop the concept of platformic torque to theorize how aesthetic, temporal, and regulatory pressures are organized through platform infrastructures and experienced differently across intersectional positions. I then analyze how resellers respond through automation, metric manipulation, rule negotiation, and fragile forms of collectivity.

Bridging critical HCI, feminist political economy, decolonial thought, and discard studies, this dissertation reframes online resale as a diagnostic site for understanding how platforms, surplus, and intersectionally differentiated labor are entangled within imperial formations of labor, consumption, and discard. Rather than romanticizing resistance, it shows how survival, complicity, and contestation coexist in secondhand markets, and how collective futures remain both imaginable and structurally fragile under colonial racial capitalism.