CS + Math at Vanderbilt, class of 2029. I build quant and ML systems.
neural-options-lab (live demo): deep-learning pricer for arithmetic Asian options. 1.33 bp RMSE against 200,000-path Monte Carlo references, rough-volatility calibration to live SPY and BTC option chains, and a neural pricing map that refits a 1,284-quote surface in seconds on a laptop CPU.
hft-lob: lock-free C++20 order book for NASDAQ ITCH 5.0. 10M+ messages/second on one core, 110M+ across 23 cores via wait-free sharding, verified byte-for-byte against a sequential reference.
hft-market-maker: a 500-seed paired study of Avellaneda-Stoikov market making. Inventory risk falls 17% and max drawdown 14%, at the cost of 36% fewer fills; the model buys risk reduction with volume, not edge.
ssda-folio-pipeline and ssda-nlp-tools: the vision and record-linkage stages that turn a 750,527-image colonial archive into a queryable database of people, events, and family ties. Sole developer, for Vanderbilt's Slave Societies Digital Archive.
Also built climateharbor.org, a climate-data platform with users in 35 countries.
LinkedIn |