I build things with AI, data, and an unreasonable number of browser tabs.
Currently exploring AI agents, ML systems, and full-stack AI products — I like turning messy problems into things that actually work.
🎓 Columbia '26 · Tsinghua '25
📍 New York · San Francisco Bay Area
🐱 The profile picture is a fairly accurate representation of my debugging process.
A reliable execution engine for turning high-level tasks into deterministic UI actions, with logical locators, replay, validation, and safety controls.
TypeScript · Playwright · Vitest
An AI-powered daily briefing platform that helps software engineers keep up with AI without spending hours reading the news.
Next.js · FastAPI · PostgreSQL · OpenAI
An end-to-end ML framework exploring Transformer-based sequential behavior modeling, gradient boosting, and ensemble learning for credit risk.
Python · PyTorch · LightGBM · XGBoost
🌾 Kaggriculture Agent — Building an autonomous strategy agent for Kaggle's farming simulation challenge.
📊 Kaggle Playground — Experimenting with feature engineering, gradient boosting, and model ensembling on tabular ML problems.
Languages: Python · TypeScript · SQL
AI / ML: PyTorch · LightGBM · XGBoost · scikit-learn
Backend & Data: FastAPI · PostgreSQL · SQLAlchemy
Frontend: Next.js · React · Tailwind CSS
Tools: Git · Docker · Playwright · GitHub Actions
Building, breaking, learning, repeat.
