Summary
Kai Zhang is a Senior Machine Learning Engineer based in Berkeley with nine years of experience building production-grade LLMs, recommendation systems, and safety-focused NLP solutions across Amazon, DiDi, and national labs. He combines deep research roots (PhD in Astrophysics) with pragmatic engineering—designing scalable LLM pretraining/post-training pipelines, RL and SFT workflows, and evaluation frameworks that drove measurable business impact like 30–48% revenue lifts and dramatic acceptance-rate improvements. Kai has led projects from safety interventions that reduced criminal rates to conversational agents and multimarket recommendation systems, often improving throughput or accuracy by an order of magnitude. Comfortable across cloud platforms, MLOps, and full-stack tooling, he also brings a knack for automating evaluators and simulators that make model metrics actionable in real environments. Notably, his transition from astrophysics to applied ML produced high-impact tooling and published science, reflecting a blend of rigorous analysis and product-oriented delivery.
9 years of coding experience
13 years of employment as a software developer
Ph.D Astrophysics, Ph.D Astrophysics at University of Science and Technology of China
Bootcamp Full Stack Web Deveolper, Bootcamp Full Stack Web Deveolper at UC Berkeley Extension