Xiaochen Zhang

Staff Machine Learning Engineer @Firework

Vancouver, BC, CA

Built & Lead in-house realtime digital human experience.* With unique avatar syncing AIGC model trained in-house.* That scales cheaply and wonderfully with many WebRTC frameworks * I Can create conversational digital human service with our own model from scratch and runs on laptop, with hundreds of ms of time to first frame/bytes. Knows well the the entirety on how each components connect for STT->LLM->TTS-> Digital Human etc.* Integrate racing LLM requests as enhanced communication experience beyond digital human (real human conversation).* Though LLM is easy and everywhere, I built and hack it into very limited latency during realtime application. Stateful LLM, RAG, conversation-flow control all as fast as possible.* Integrated AIGC models into helm x Kubernetes for infrastructure.The go to engineer for company to try something uncanny and impossible.

Northern Arizona University

Master of Science - MS, Computer Science

East China University of Political Science and Law

Bachelor's degree, International Economics