CMU CS '29. Interested in ML, quantitative research, and systems that make careful use of real data.
Background: USAMO qualifier · USACO Platinum · SAT 1600/ACT 36 · ML Intern @ Cartesia
- polymarket-resolution-drift — measured how fast Polymarket prices converge to the correct outcome before resolution, and backtested a late-game correction strategy against realistic spread/slippage assumptions (real data)
- lichess-opening-divergence — statistically rigorous analysis of how the empirically best chess opening move shifts across rating bands, using bootstrapped confidence intervals, multiple-comparison correction, and time-based out-of-sample validation.
Submitted fixes to real, verified bugs in:
- scikit-learn — Pipeline HTML repr losing step names
- Netflix Metaflow — card component breaking on NaN/Infinity values
- Palantir baseline-error-prone — incorrect SuggestedFix for StringBuilder chains
- Meta Encodec — channel mismatch causing a RuntimeError
- Google Guava — ArrayStoreException edge case in ObjectArrays.concat
Python Java C++ · PyTorch TensorFlow scikit-learn NumPy pandas OpenCV · Git LaTeX