I’ve lead various efforts in perception to improve advanced vehicle understanding, such as:

• E2E development of multi-task camera/lidar ML models

• Model consolidation for onboard/offboard optimization

• System-wide improvements for super long-tail action recognition (i.e. avoiding collisions)

• Multi-frame foundation models & their applications

• ML flywheel to enable rapid scaling in new cities

I’ve collaborated with many folks in adjacent teams (Product, Planner, Eval, etc.), and also authored several patents.

Pony.ai is an international AV company that has driven over 20 million miles, with testing sites located in the US, China, Luxembourg, Korea, and Saudi Arabia. Pony.ai IPO'd on NASDAQ in Nov 2024.

I was an engineer in the Fremont office, and developed all of the camera-based ML capabilities (both onboard/offboard systems).

10th hire at YC-backed startup (Winter 2015). Built automated checkout solutions using cutting-edge ML techniques, and on-device transaction processing services. Last valued at ~1.5B in Series B funding (May 2022).