Summary
Yi Pan is an engineering leader with 12 years of experience building high-performance teams and production-grade AI infrastructure, currently managing Rufus at Amazon. He has deep technical expertise in conversational AI, LLM agents, RAG, SFT and RLHF, and has architected scalable training frameworks for 100B+ dense and MoE models, improving PPO throughput by 5x. Yi has a proven track record of founding engineering organizations from 0-to-1—leading cross-functional teams of 50+ globally to deliver multimodal assistants, dialog management services, and image-relevance improvements that raised CSAT. He combines hands-on optimization of training pipelines on GPU and Trainium clusters with an impact- and data-driven agile culture. Trained originally as an archaeologist (M.A. Anthropology), he brings a rare mix of rigorous research thinking and product-focused engineering to complex AI alignment problems.
12 years of coding experience
11 years of employment as a software developer
Master of Arts (M.A.) Anthropology (Archaeology), Master of Arts (M.A.) Anthropology (Archaeology) at University of Washington
Bachelor of Arts (B.A.) Archaeology, Bachelor of Arts (B.A.) Archaeology at Sichuan University
Chinese, English