projects
Selected research and projects. Papers available upon request.

Vector Balancing Games
MATH 159 (Discrete Probabilistic Methods), 2026
details
Studied Alon & Spencer's vector balancing games. Introduced vector-balancing models for online discrepancy minimization. Explored constructive algorithms for online & offline settings and balancing game variants.

Black Box to Bedside: Distilling Reinforcement Learning for Sepsis Time Series
NeurIPS TS4H & CogInterp Workshops, 2025
details
Trained deep Q-network on sepsis treatment data, augmenting with ensembles & regularization. Distilled into interpretable decision trees while maintaining >85% clinician agreement. Awarded CS224R outstanding project (top 6 out of 650+ graduate & undergraduate students).

VeriBench: End-to-End Formal Verification Benchmark for Code Generation
ICML AI4Math Workshop, 2025
details
Developed benchmark of 100+ examples for evaluating language model capabilities in full Python-to-Lean code generation. Worked on self-optimizing Trace agent architecture for improved formal verification. Nominated for workshop oral presentation (top 1).

Heap Allocator
CS 107 (Computer Organization & Systems), 2025
details
Designed & implemented full implicit & explicit free list heap allocator supporting coalescing, splitting, in-place realloc. Built custom debugging utilities. Achieved >80% utilization on real allocation traces.