I am a final-year PhD candidate in Machine Learning in the H. Milton Stewart School of Industrial & Systems Engineering (ISYE) at Georgia Institute of Technology (GT), advised by Dr. Gian-Gabriel Garcia and Dr. Kamran Paynabar.
About My Research
My research program treats data as a strategic resource rather than merely an information flow, which enables me to create integrated frameworks that merge optimization, statistical modeling, and AI for operational decision-making. This data-centric operations research perspective has led to significant methodological contributions across public health operations and personalized medicine. By developing novel models that explicitly leverage domain-specific data engineering techniques to integrate predictive analytics with prescriptive decision-making, my work provides actionable insights that improve both the efficiency and effectiveness of complex service operations.
My methodological contributions span several key areas of operations research and operations management. I have developed novel stochastic process networks for spatio-temporal prediction and resource allocation, Bayesian compartmental models for sequential decision-making under incomplete surveillance, and doubly data-driven distributionally robust optimization frameworks that integrate risk-scoring tools with prescriptive treatment planning. More recently, I have been working on post-training alignment of Large Language Models and Foundation Models. I have developed constraint-aware approaches for safely integrating large language models into operational decision-making, addressing critical concerns about AI safety in high-stakes operations. Moreover, I am developing methods for foundation model knowledge distillation for early disease outbreak control. These methods advance operations research theory while maintaining strong practical relevance through extensive collaborations with public health agencies, hospital systems, and healthcare organizations.
- Public Health Operations: Considering delayed surveillance, incomplete data, and epidemic dynamics, how can we design mechanism-inspired statistical models and optimal data acquisition/sharing policies to enhance disease monitoring (e.g., opioid overdose, COVID-19) and provide actionable insights for multi-scale public health interventions?
- Personalized Medicine: Considering dataset shifts, performance disparities, and uncertainties in AI/ML risk-scoring tools, how can we develop distributionally robust algorithms that augment clinicians’ treatment planning while ensuring fairness and adapting to evolving patient populations through performative prediction frameworks?
- Foundational Data Science with Post-Training Foundation Model: What is the new paradigm of data-driven optimization and process control with AI, e.g., Foundation Models and Large Language Models?
News
I will present my Job Market Paper on Constraint-Aware Self-Improving Large Language Model for Clinical Role Model Generation in INFORMS 2025. Preprint available at: Here
- Session title: SD 42 Al & ML Applications in Healthcare
- Time: Sunday October 26th 3:00 PM — 3:15 PM.
- Location: Building B Level 3 B302