New York City Metropolitan Area
Coding RL Bigrun + RL Science team: core contributor to Grok 4.5 Previously worked on Forecasting: achieved SOTA
Buy low sell high and research new strategies to do so.
Researched and enhanced LLMs ability to generate correct code for competitive programming problems
Found stronger mathematical guarantees for utility functions that generate an S-shaped convexity in the positive outcome of a weighted flip after paying the theoretical fair bid price. Constructed counterexamples that proved previous conditions for S-shaped convexities were too weak (necessary but not sufficient)
Quantified robustness of Fourier Neural Operators for PDEs by adding pixel-wise noise during testing. Improved robustness through data augmentation, noise training, and stability training. Decreased test error and enhanced robustness of FNO on the 2D steady-state Darcy Flow dataset by 35%.
Developed ML method extending on mixed-effects random forests and fuzzy forests to deal with longitudinal high-dimensional data that are both highly correlated through time and data features.
Worked on the Equities (StatArb) and Options desks, modeling future volume, customer flow, and returns using TRF data, and built forward volatility term structure curves for commodity futures.