Joseph E. (Joey) Gonzalez · Professor of EECS, UC Berkeley
Conference organizers, hosts and journalists are welcome to use any of these biographies, edited for length as needed. A headshot and CV are also available.
Joseph E. Gonzalez is a professor in EECS at UC Berkeley working at the intersection of machine learning and computer systems. With his students and collaborators, he has worked on widely used systems including GraphLab, Ray Serve, Ray Tune, vLLM, SGLang, Chatbot Arena, Gorilla and MemGPT. His current research spans AI agents, post-training and new neural architectures.
Joseph E. Gonzalez is a professor in EECS at UC Berkeley whose research spans machine learning and computer systems. His work asks how new systems abstractions can enable new forms of machine learning. He has pursued that question from large-scale graph computation (GraphLab, PowerGraph, GraphX), through model serving and tuning (Ray Serve, Ray Tune), to infrastructure for large language models. His students and collaborators have contributed to widely used systems including vLLM, SGLang, Chatbot Arena, Gorilla and MemGPT. Gonzalez is a founding member of Berkeley's Sky Computing Lab, LMSYS and RISELab. His current research spans AI agents, post-training methods that let models keep learning, and new neural architectures for efficient, scalable models.
Joseph E. Gonzalez is a professor in the Department of Electrical Engineering and Computer Sciences at UC Berkeley. His research, at the intersection of machine learning and computer systems, asks how new systems abstractions can enable new forms of machine learning and AI.
As a PhD student in machine learning at Carnegie Mellon, he developed GraphLab and PowerGraph for large-scale graph computation. He co-founded Turi (formerly GraphLab), which Apple acquired. As a postdoc in Berkeley's AMPLab, he led GraphX, now part of Apache Spark. At Berkeley his group turned to machine-learning infrastructure. That work included Clipper for prediction serving and Ray Tune for hyperparameter tuning. Simon Mo developed Ray Serve from Clipper and later became a PhD student advised by Gonzalez. The group then moved to infrastructure for large language models. His former student Lianmin Zheng created SGLang together with Ying Sheng and contributed to vLLM and Chatbot Arena. Simon Mo later co-founded Inferact, the company built around vLLM. His former students Shishir Patil and Tianjun Zhang led Gorilla, and Charles Packer, Sarah Wooders and Kevin Lin created MemGPT. Gonzalez's current research spans AI agents, post-training and continual learning, and neural architecture innovation.
He is a founding member of Berkeley's Sky Computing Lab, LMSYS and RISELab. His honors include a VLDB Test of Time Award for GraphLab, the ACM SIGMOD Systems Award for Apache Spark (shared) and an NSF CAREER Award.
Gonzalez shapes his research agenda around one question: “If we solve today's hot problems, what new problems will those solutions create?” That question led him to machine-learning systems in 2008, when most researchers were developing new models. It led him to model serving and inference systems in 2015, as the field focused on scaling training. Today he is betting that AI agents will become highly capable and widespread. He studies how systems must change for thousands of such agents, how people can steer and negotiate through them, and how models can move from learning from data to learning from teachers.
