Experience

Building the LLM agent that turns a recruiter's chat into a structured hiring spec, and the backend behind candidate matching. Co-architected a two-stage recommendation system over 160K+ profiles and cut hiring-spec insights latency from 22s to 0.2s.

Built backend services for AI customer support across 3 enterprise clients and 60+ properties: an agent workflow engine, multi-tenant message ingestion, and fault-tolerant AI request handling. Processed 11,800+ tickets and raised AI reply success from 83% to 96% under 200+ concurrent requests.

Built perception algorithms for a robotic ultrasound system, fine-tuning AlphaPose to guide robot-arm positioning and cutting image origin-detection latency from 200 ms to 11 ms.