Quantum computers can solve important computational problems in many application domains such as chemistry, material science, high-energy physics, cryptanalysis, and machine learning. Most of these applications are intractable on conventional systems. Quantum computers get their computational advantages by leveraging quantum-mechanical properties to store and manipulate information. A quantum bit or \textit{qubit} is the fundamental unit of information on a quantum computer. Quantum algorithms manipulate the state of the qubits using quantum operations. After years of research and development, quantum computers with few hundreds of qubits are available today. Unfortunately, the qubit devices are noisy and imperfections in the quantum operations lead to incorrect outcomes during program execution and limit the fidelity of these systems. Quantum information can be protected by using quantum error correction (QEC) codes at the expense of redundancy (50-1000x). These codes project errors into failed parity checks which are used to identify or decode errors in real-time. However, it is impractical to run applications in a fully fault-tolerant manner on emerging systems with only few hundreds to thousands of qubits. Instead, these systems run applications in the presence of errors and promise to accelerate certain domain-specific applications. Quantum hardware errors serve as a major bottleneck in running most practical quantum applications and their impact must be minimized. %and therefore, heavily rely on error mitigation techniques that improves the quality of the solutions. Thesis Statement: By using software techniques to mitigate the impact of specific types of qubit errors and architecture solutions to detect errors in real-time, we can improve the fidelity of emerging quantum computers.