Date: 04/20/2026

Time: 11:30 AM

Location: Groseclose 304 & https://gatech.zoom.us/j/97709876945?pwd=QJxr1oKt9QYGHNcShdytZbC6hrhjb8.1

Student Name: Che-Yi Liao

Machine Learning PhD Student

Home School: ISYE
Georgia Institute of Technology

Committee

1 Gian-Gabriel Garcia (Advisor)

2 Kamran Paynabar (Advisor)

3 Yao Xie

4 Jing Li

5 Esmaeil Keyvanshokooh

Abstract

This dissertation investigates how artificial intelligence systems can be operationalized through data-centric design so as to remain reliable, interpretable, and operationally meaningful in real decision environments. It advances this thesis through three research streams: data-centric AI systems across the application domains of public health and personalized medicine, and foundational data science for AI systems. Across these streams, the dissertation introduces new methodologies that integrate prediction, optimization, and action in the presence of delayed reporting, incomplete observations, local data limitations, and distributional instability. Collectively, this work argues that the operationalization of AI is fundamentally a data-centric design problem.