SF Hackathon
Taught a robot arm to pour latte art through physical demonstration at a hackathon in San Francisco. Learning from demonstration pipeline for precise fluid pouring motions.
~90% less manual tuning
Autonomous ML agent built at the OpenAI NYC Hackathon. Trains, evaluates, and compares multiple models end-to-end with a retraining module that improves baseline accuracy by up to 12%.
Real-time verification
Coordinated multi-agent system where AI agents debate claims using shared context and role-based reasoning. Real-time fact checking and source verification across agent responses.
~60% faster response
Real-time conversational AI on a small embedded device (ESP32) that generates personalized spoken stories. Fast response pipeline delivering spoken output in under 3 seconds.
70% less manual input
iOS app for voice-based food expiration tracking using speech recognition. Reduces manual interaction for pantry management.
Published on npm
Open-source CLI tool for parameterized JSON mocks with latency injection. Published on npm for developer workflow optimization.