I'm an ML engineer who likes building things and understanding why they work. This site is a collection of projects I've built to feed that curiosity. Some to sharpen my understanding of a topic, others just because they sounded fun to make.
I studied math at Northwestern and computational math at Stanford. After a few years working in San Francisco, I moved to Paris to learn French and experience life in Europe. I'm now back home in Texas after more than a decade away.
Outside of work, I like hiking, cooking, playing tennis, and getting my heart broken by the underperformance of Texas A&M football.
Agentic RAG system that answers questions about US presidents, adapting its search strategy per question. Responses are grounded in a 500+ article knowledge base and cite their sources. Tuned with systematic evals over the orchestrator LLM, retrieval/reranker models, and search depth.
AI agent that autonomously plays Wikipedia Golf: the game of navigating from one page to another using the fewest links possible.
Generative language models fine-tuned on the works of Dostoevsky, Fitzgerald, and Twain to capture their distinctive writing styles.
My attempt at re-creating Google Translate from scratch. English/French translation using encoder-decoder transformers.
A conversational AI fine-tuned on podcast transcripts to capture the distinctive speaking styles of famous hosts.
Convex optimization solver built from scratch using only primitive linear algebra operations.
Custom method for ranking college football teams inspired by Google’s PageRank algorithm.
Visualizes a Byte Pair Encoding (BPE) tokenizer built from scratch.
Web app that extracts Sudoku puzzles from uploaded photos and solves them with a backtracking algorithm.
