A portfolio of quantitative trading systems, machine learning models, and fullstack applications that demonstrate professional-grade engineering and measurable impact.
Python, Polars, DuckDB, Parquet, Anthropic API
An AI product for commercial-auto claims reconciliation. It reads adjuster notes on open claims and flags the ones where a documented fact is not reflected in the carried reserve. It is a reconciliation tool rather than a prediction machine.
Python, pandas, IBKR API, Supabase
The quantitative engine behind breakouts.trade: a channel-breakout strategy implemented as a single source of truth shared by research and execution, a nightly scanner pipeline over a local market-data lake, and a live Interactive Brokers trading robot that imports the same strategy code through a parity bridge.
Python, pandas, Jupyter
A deterministic, rule-based strategy that trades early-morning momentum divergence between leveraged ETFs. It opens equal-dollar positions in UPRO and SPXU at the market open, closes the losing side once a spread threshold resolves within a five-minute evaluation window, and exits the winner on a configurable schedule.
Econometrics coursework project
Quantile-regression study testing whether pre-breakout volume contraction predicts upper-tail returns on Qullamaggie-style momentum setups. UC Berkeley EC143 final project.
Python, CodeCarbon, pandas, Jupyter
A measurement pipeline for the energy cost of LLM API usage. It ingests and cleans large conversation datasets, replays the extracted prompts against several hosted models, and records power draw with CodeCarbon while engineering 30+ linguistic features per prompt.
Next.js, TypeScript, React, Python, PostgreSQL
Built a Duolingo-style platform to learn a breakout stock-trading strategy with real charts and simulations. Shipped a production product with 458 signups across six continents through iteration with Reddit and X.com communities.
React, React Router, Tailwind CSS
A React application built for a regional environmental organization, with keyword search and filtering to help visitors discover member groups by location and type, an interactive image gallery, and a responsive Tailwind layout.
Next.js, React, Tailwind CSS, Contentful
This site: a Next.js App Router portfolio and blog. Posts are authored either as local MDX or in Contentful, rendered through a shared pipeline, with JSON-LD structured data and a generated sitemap for search engines.
Open to projects, research collaborations, and innovative work in AI, finance, and technology. If you're building something that matters, I want to hear about it.