Projects

Engineered an autonomous arbitrage bot exploiting price inefficiencies across Kalshi, Polymarket, and Augur by anchoring to Kalshi's liquidity and executing counter-trades on lagging, lower-volume markets. Captured risk-free spreads amid fragmented, immature prediction markets. Made money from the election :)

Youngest GillingsX TED presenter. Built the first AI-driven youth mental health platform integrating 5,000+ scraped and open-sourced data points. Visualizes disparities via interactive heatmaps, combining prevalence, access, and outcome data across gender, race, and geography.

Built a Playwright-based Python scraper to automate NC DMV appointment bookings for securing in-state tuition. Extracted booking volume data to infer site traffic and appointment demand trends. Approach generalizes to other reservation systems for semi-reliable web traffic analytics.

Founder of Crosstime, an MIT open-source scheduling platform used by hundreds at UNC's Kenan-Flagler. Enables timezone-aware group coordination with a cleaner, more responsive interface than When2Meet.

Led development of the largest scalable U.S. mental health provider database. Built custom web scrapers from semantic metadata Google, aggregating 100k+ providers. Fully open, no-cost, and login-free access.

Volunteer-led review sessions for Calculus I–III students at UNC Chapel Hill. I run conceptual deep dives followed by engineered solutions using real-world applications. To help students bridge theory and practice, I code solutions in C++, OCaml, and Python—often pulling from fields like quantitative finance, AI, and data science.

Led development of an incremental learning portfolio optimization project using mean variance optimization, Markowitz's Efficient Frontier, and Hidden Markov Chains under PI Dr. Ivan Cherednik. Presented at UNC Chapel Hill's Undergraduate Research Symposium in April 2024.

Built personal website with realtime Spotify cards, arduino-based computer vision for bookshelf view, AWS CDN integration, reverse-engineered iCloud components, and numerous AI/ML easter eggs.

Founded Contour Developments, Inc. to enhance blockchain security and verification. Built high-frequency supervised feeds and migrated banks to Ethereum using Solidity smart contracts and a Python-driven backbone.

I developed AI models during an independent study under Mr. Eric Dirks at Mayo High School. I built a GPT-3.5 detection system using RoBERTa and Bayesian analysis. I designed a Bluetooth-only network tested over 1500 ft in 3D space and a PyTorch HMM to convert mouth movements into subtitles. My final project trained a generative language model in PyTorch with self-curated data to match early GPT-2 performance.

Worked with team of 4 (Cooper Gamble, Benjamin Erickson, Marc Zoghby, and Phillip Wisnewski) to top 19% among 650 submissions. Developed mathematical models to analyze global e-bike sales, assessing impact on carbon emissions, traffic congestion, and health, quantifying effects on reducing use of other transportation modes.

Founded and developed a CLI-based application used by 250+ users to automate buying sneakers. Reverse-engineered APIs, websites, and mobile apps of multi-billion dollar companies like Target, Yeezysupply, Footlocker, Nike, and many more. Open-sourced the first ReCAPTCHA v2 computer vision (PyTorch) solver. Wrote scalable APIs to handle 100,000+ requests per second.