I am currently a first year Ph.D. student in the Theory group at the University of Southern California fortunate to be advised by Shaddin Dughmi and Shang-Hua Teng, while also working closely with Haipeng Luo. Previously, I was a Research Assistant in the Machine Learning Theory Group at Cornell University and prior to this, I recieved my M.S. in Computer Science from Cornell University (2024) where I was advised by Karthik Sridharan. I recieved my B.A. in Computer Science and Mathematics from Cornell University (2022) under the supervision of Noah Stevens-Davidowitz.

Research Interests

Currently I’m interested in avenues for developing a more practical theory for machine learning. This has taken shape in my current research with projects aiming to answer the following questions:

• Learning with Auxiliary Information: Can auxiliary information—such as chain-of-thought trajectories, unlabeled data, distributional structure, or other forms of side information—fundamentally alter the statistical complexity of learning? In particular, can it turn tasks that are provably unlearnable from labeled input–output data alone into learnable ones, while also reducing sample complexity or accelerating learning in settings that are already learnable?

• The Power of Hallucination in Language Generation: Is hallucination a necessary condition for creative breadth? When can data-driven hallucination become a productive form of imagination, and how can we ensure it remains benign?

• Adaptive Algorithms: Can we build algorithms for interactive learning that are capable of achieving tighter regret bounds under favorable conditions while still maintaining optimal guarantees in the worst case?

• Relaxing Realizability: Are there practical and interpretable conditions that ensure learnability of regression-based learning paradigms—such as active learning, bandits, and reinforcement learning—beyond the classical realizability assumption?

More broadly, I am interested in uncovering the theoretical foundations of machine learning.