Builders & open source
Joseph E. Gonzalez
Associate Professor, Department of Electrical Engineering and Computer Sciences, UC Berkeley
- Associate Professor, Department of Electrical Engineering and Computer Sciences, University of California, Berkeley21
- Founding member and advisor, LMSYS25
Joseph E. Gonzalez is an associate professor in EECS at UC Berkeley working at the intersection of machine learning and computer systems, and a founding member of the Sky Computing Lab, LMSYS and RISELab. With students and collaborators, Gonzalez has worked on systems including Gorilla and the Berkeley Function Calling Leaderboard (BFCL), Chatbot Arena, SGLang, MemGPT, vLLM, Ray Serve, Ray Tune and GraphLab. Gonzalez is a co-author of the Gorilla and vLLM papers and an advisor to LMSYS; current research spans AI agents, post-training and new neural architectures.12345
At a glance
- Current roles
- Associate Professor, Department of Electrical Engineering and Computer Sciences +1 more
- Companies
- UC Berkeley, LMSYS
- Sector
- University
- Education
- Carnegie Mellon, UC Berkeley
- Work listed
- 2
- Honors
- 3
- Sources cited
- 9
- Profile reviewed
Career
Joseph E. Gonzalez, also known as Joey Gonzalez, has been on the UC Berkeley EECS faculty since 2016 and is a member of Berkeley AI Research (BAIR). As a PhD student in machine learning at Carnegie Mellon, advised by Carlos Guestrin, Gonzalez developed GraphLab and PowerGraph for large-scale graph computation, and co-founded Turi (formerly GraphLab), which Apple acquired in 2016.26
As a postdoc in Berkeley's AMPLab, Gonzalez led GraphX, which became the graph library of Apache Spark. Gonzalez is a founding member of Berkeley's RISELab, the Sky Computing Lab and LMSYS, and is listed among LMSYS's advisors.265
Research
Gonzalez's group studies how new systems abstractions can enable new forms of machine learning. After graph computation, the group turned in 2015 to serving and tuning models, with Velox, Clipper for low-latency prediction serving, InferLine for serving model pipelines and Ray Tune for hyperparameter tuning; Simon Mo developed Ray Serve from Clipper.62
From 2021 the group worked on infrastructure for large language models, including Alpa and AlpaServe, the vLLM paper on PagedAttention memory management for LLM serving, SGLang, Chatbot Arena and the Gorilla project on connecting LLMs to APIs. Since 2023 the group's research has covered AI agents, post-training and continual learning, and new neural architectures, with projects such as MemGPT and SkyRL.6437
Open-source work
Several systems from Gonzalez's group are released as open-source software. The Gorilla repository, which hosts the Berkeley Function Calling Leaderboard, is under the Apache-2.0 license, as is the vLLM repository. Gonzalez's biography credits former students and collaborators with leading these systems, among them Shishir Patil and Tianjun Zhang for Gorilla and Lianmin Zheng and Ying Sheng for SGLang.8972
Work and contributions
Gorilla and the Berkeley Function Calling Leaderboard (BFCL) Open-source code
Co-author of "Gorilla: Large Language Model Connected with Massive APIs" (2023). Gonzalez's research page says Gorilla was led by Gonzalez's students Shishir Patil and Tianjun Zhang and that BFCL is now led by Gonzalez's student Huanzhi Mao; the Gorilla repository, which holds the BFCL code and data, is Apache-2.0 licensed.3678
vLLM Open-source code
Co-author of "Efficient Memory Management for Large Language Model Serving with PagedAttention" (2023), the vLLM paper. The vllm-project/vllm repository is Apache-2.0 licensed.469
Timeline
- Published the first GraphLab paper, at UAI 2010, as part of doctoral work at Carnegie Mellon.6
- Published Distributed GraphLab at VLDB 2012 and PowerGraph at OSDI 2012.6
- Published GraphX, graph processing inside a distributed dataflow framework, at OSDI 2014.6
- Joined the UC Berkeley EECS faculty; Turi, which Gonzalez co-founded, was acquired by Apple the same year.26
- Published Clipper, a general-purpose, low-latency prediction serving system, at NSDI 2017.6
- Received the VLDB 2023 Test of Time Award for Distributed GraphLab.6
- Co-authored the Gorilla paper on large language models connected with massive APIs.3
- Co-authored the vLLM paper on PagedAttention memory management for LLM serving.4
- Co-authored the Chatbot Arena paper at ICML 2024.6
Other documented roles
Roles the sources document that are not confirmed as current.
Honors and recognition
Where they work
- UC Berkeley 27 people in the wiki
- LMSYS 3 people in the wiki
Related people
- Andy KonwinskiAlso at UC Berkeley
- Ion StoicaAlso at UC Berkeley
- Hao ZhangAlso at UC Berkeley
- Shreya ShankarAlso at UC Berkeley
- Owain EvansAlso at UC Berkeley
- John YangAgents · Evaluation and benchmarks
- Mike KnoopEvaluation and benchmarks · Agents
- Greg KamradtEvaluation and benchmarks · Agents
- Gabe PereyraAgents · Evaluation and benchmarks
- Thomas ScialomAgents · Evaluation and benchmarks
Suggest a correction or removal
Profiles cover public professional work only, and each statement cites a source. If something is out of date, missing or wrong, tell us and link a source. People can also ask for their profile to be corrected or removed.
References
9 sources · profile reviewed Oct 21, 2026 · data updated Oct 21, 2026
- Joseph E. Gonzalez (Joey) — Gonzalez Group at UC Berkeley
- Biography · Joseph E. Gonzalez, UC Berkeley
- Gorilla: Large Language Model Connected with Massive APIs (arXiv 2305.15334)
- Efficient Memory Management for Large Language Model Serving with PagedAttention (arXiv 2309.06180)
- About LMSYS (officers and advisors)
- Research Through the Years — Gonzalez Group
- Berkeley Function Calling Leaderboard (BFCL) V4
- ShishirPatil/gorilla repository (Apache-2.0 license)
- vllm-project/vllm repository (Apache-2.0 license)