Summary
Dongxu Zhang is a Lead AI Scientist with 11 years of experience building and shipping LLM-based agents and safety systems, currently leading AI work at Optum. He earned a Ph.D. in Computer Science from UMass Amherst under Andrew McCallum and previously spearheaded modeling and safety guardrails for ASAPP’s Generative Agent product. His work spans generative AI agents, AI safety, synthetic data generation and evaluation, and optimization of LLMs, with practical experience fine-tuning models for PII redaction and hallucination detection (ACL 2024). Dongxu blends deep research with production engineering—designing structured reasoning mechanisms, function-call formatting, and end-to-end user-simulation evaluations. He also has a strong background in knowledge representation, relation inference, and large-scale information extraction from internships at Google and Amazon and research roles in top universities. Based in Atlanta, he combines rigorous academic foundations with hands-on product impact in applied AI safety and agent design.
10 years of coding experience
4 years of employment as a software developer
Master of Science - MS Information and communication engineering, Master of Science - MS Information and communication engineering at Beijing University of Posts and Telecommunications
Doctor of Philosophy - PhD Computer Science, Doctor of Philosophy - PhD Computer Science at University of Massachusetts Amherst
Chinese, English