Currently running 5 autonomous AI agents 24/7 on a Mac Mini. They ship code, trade options, analyze biotech, and learn from their mistakes.
PhD Economics, UCLA · BA Statistics & Economics, UC Berkeley
我們能做的，就是將自己的事做到最好，其它，交給命運吧。
All we can do is do our own work to the best of our ability. The rest, leave to fate. — 顧鳴高
A neuron-inspired memory graph for AI agents where nodes fire like neurons and edge weights learn from outcomes via STDP. Built from running three agents for 20 days and spending $13K on wasted context, with v12.2.0 adding policy gradients and self-learning.
→ Read the page · Policy gradient derivation · GitHub · v12.2.0 Blog Post
A high-trust AI agent running 24/7 on a headless Mac Mini via OpenClaw. Manages code deployments, monitors systems, triages emails, and coordinates two sub-agents through Telegram.
→ Runs three agents across two products with shared learning
A web app for neighbors to share homegrown produce. Built with React 19, Express, Postgres, and Cloudflare, with an AI agent managing coding, testing, and deployment workflows.
→ Live at bountiful.garden · AI-assisted development
ML-driven options trading system using GBM + TFT ensemble models for credit spread selection. Handles ingestion, feature engineering, training, calibration, and autonomous execution in one pipeline.
→ End-to-end autonomous options pipeline
An always-on AI analyst built on OpenClaw for a biotech-focused asset manager. Ingests feeds, triages trial events, tracks FDA catalysts, and writes pre-market briefings.
→ 20+ automated crons · Real-time intelligence pipeline
Machine Learning Engineer