Projects
A novel approach to options contract pricing, built on a dataset of 200M+ contracts over 17 years. Achieved 7.9% improvement in MAE by encoding daily options volatility surfaces into RGB images, using a CNN to extract a context vector, and passing through a causal self-attention transformer.
Machine learning classification of particle precipitation events across 6+ years of multi-dimensional data from the twin ELFIN satellites. Built custom Julia and Python data libraries to train a YOLOv8x model on university HPC clusters, reducing false positives by ~80%.
Live market-making across 12 stablecoin instruments on Kraken US and OKX US. CU Boulder's student trading team maintained a 98.44% uptime across a 1–178 bps spread capture range, achieving 137.63% annualized returns.
An empirical study on the performance cost of abstraction in Java object-oriented systems. Explores performance benchmarks on various JVM configurations across abstraction levels using AWS computing and JIT performance analysis.
A personal portfolio (this website!) built with React, Next.js, and Tailwind CSS, deployed on Vercel. An "interactive resume" built with modern code practices and a trial of AI-assisted coding on familiar territory from prior internship experiences.