I build frameworks and tools that empower developers to create agentic AI applications. Currently I'm the lead engineer and principal architect for the Microsoft Agent Framework. Previously I led Semantic Kernel to GA, developed autonomous driving algorithms at Xpeng Motors, and published reinforcement learning research at IROS 2019.
Lead engineer and principal architect for the Microsoft Agent Framework. Designed agentic workflow system with Pregel-style runtime, human-in-the-loop execution, checkpoint, and end-to-end observability.
Led Semantic Kernel to General Availability, scaling to 170M+ monthly Azure OpenAI requests. Contributed to OpenTelemetry GenAI Semantic Conventions.
Built end-to-end pipeline for rapid prototyping of custom computer vision models, reducing development cycles from months to under one week.
Designed motion prediction algorithms for dynamic traffic environments. Deployed models to production autonomous vehicles.
Developed multi-object tracking algorithms using radar-camera fusion. Achieved 150m detection range on production vehicles.
Led reinforcement learning research for quadrotor control. Built simulation environments and neural network pipelines for real-time flight control.
Developed telemetry protocol for the bi-pedal robot Cassie.
Best Research Award 2019
Honor Roll · Engineering Scholarship