Research

My research develops theoretical foundations and scalable algorithms for reliable, data-efficient reinforcement learning and control in large-scale networked systems. I study how imperfect but useful structure—including coarse predictions, decomposable dynamics, optimization geometry, stability, and operational constraints—can reduce the information and computation required for learning. A central goal is to establish rigorous guarantees on sample complexity, near-optimal performance, stability, and constraint satisfaction when data are limited and failures are costly. I apply these ideas to power systems and grid-aware AI computing, where online decisions must coordinate physical and computing resources under uncertainty.