PeopleofSuperintelligence

Builder

Siyuan Zhuang

Also known as 庄思源, Ryans, Ryan, SIY.Z, suquark

Siyuan Zhuang is an AI systems builder whose work spans distributed computing, language-model serving, and evaluation. He began computer science studies at USTC in 2014, earned a bachelor's degree through its Honors Program, and completed a UC Berkeley PhD in 2024 with Ion Stoica and Dawn Song as advisers. His Berkeley thesis presents Hoplite and ExoFlow for fault-tolerant distributed systems. His other work includes PagedAttention-based vLLM, the Vicuna chatbot and LLM evaluation, and Starburst, whose artifact received a USENIX Distinguished Artifact Award in 2024. He coauthored a study on budget-aware test-time scaling in 2025, and since 2026 he is Tech Lead of AI at SAIR.

Now at
SAIR
Works
15
Badges
Links
Google Scholar, LinkedIn and GitHub

Publication numbers from OpenAlex, as of 22 October 2026. Citations and h-index are hidden because this OpenAlex record may be incomplete or mix in other authors.

Timeline

  1. 2014–2018

    Computer science at USTC

    He begins undergraduate computer science studies at USTC in 2014 and earns a bachelor's degree in computer science through its Honors Program.

  2. 2019–2024

    A Berkeley PhD

    From 2019 to 2024, he works as a Graduate Student Researcher at UC Berkeley and a PhD student at the International Computer Science Institute. He completes a computer science PhD at Berkeley in 2024 with Ion Stoica and Dawn Song as advisers; his thesis presents Hoplite and ExoFlow for fault-tolerant distributed applications.

  3. 2021

    Hoplite for distributed systems

    In 2021, he coauthors Hoplite, an efficient and fault-tolerant collective-communication layer for task-based distributed systems.

  4. 2023

    Building vLLM with PagedAttention

    In 2023, he coauthors the PagedAttention paper and helps build vLLM, an LLM-serving system based on the algorithm. He contributes to Vicuna, an open-source chatbot; the FastChat-vLLM integration serves it to millions of users.

  5. 2023

    Judging language models

    He coauthors the 2023 paper “Judging LLM-as-a-judge with MT-Bench and Chatbot Arena.” It verifies agreement between LLM judges and human preferences through MT-Bench and Chatbot Arena.

  6. 2024

    Starburst in hybrid cloud

    In 2024, he coauthors Starburst, a hybrid-cloud scheduler that adjusts job waiting times to improve cluster use and balance cloud costs against completion times. Its artifact receives the Distinguished Artifact Award at USENIX ATC.

  7. 2025

    Budget-aware test-time scaling

    In 2025, he coauthors “Budget-aware Test-time Scaling via Discriminative Verification,” which combines discriminative verifiers with self-consistency to improve LLM reasoning under a fixed compute budget.

  8. Since 2026

    Tech Lead of AI at SAIR

    He is Tech Lead of AI at SAIR, where his work focuses on machine-learning systems, distributed computing, and LLM evaluation.

Career

  1. Since 2026SAIRTech Lead of AI
  2. 2019–2024University of California, BerkeleyGraduate Student Researcher

Education

  1. 2024University of California, BerkeleyPhD in Computer Science
  2. 2024International Computer Science InstitutePhD student
  3. From 2014University of Science and Technology of ChinaBachelor's degree in Computer Science

From the biography.

  1. More people shaping AI studied at Massachusetts Institute of Technology than at any other institutionIncludes University of California, Berkeley. We estimate where the people shaping AI studied and where their careers led. Massachusetts Institute of Technology leads with 467 former students, including 291 recorded doctorates.
  2. The company share of AI researchers’ top papers was 31% in 2020–2024 and 8% in 1970–1974Includes University of California, Berkeley. We estimate where the people shaping AI worked when their most-cited papers appeared: the company share was 31% in 2020–2024 and 8% in 1970–1974, and Google led all organizations in 2020–2024.
  3. The median h-index is 48 at universities and 14 at frontier AI labsIncludes University of California, Berkeley. The median h-index is 48 at universities and 14 at frontier AI labs. We compare publishing careers, recent citations, dual appointments and job moves across 2,420 university researchers and 308 frontier-lab researchers.
  4. Companies supply the most prior-work hires at 7 of the 9 frontier AI labsIncludes University of California, Berkeley. Our estimates of the organizations where 550 people worked directly before joining 9 frontier AI labs across 620 career transitions, comparing the main feeder organizations per lab.

Colleagues

People named in this biography, and people with dated positions at the same organization in the same years.

Sources and credits

Photo: We have no published photo of Siyuan Zhuang yet. Send us a free-licensed photo.

  1. Siyuan Zhuang (庄思源) suquark.github.io. Checked 21 October 2026.
  2. Siyuan Zhuang — Lacuna lacuna.tiptreesystems.com. Checked 21 October 2026.
  3. Providing Efficient Fault Tolerance in Distributed Systems | EECS at UC Berkeley www2.eecs.berkeley.edu. Checked 21 October 2026.
  4. vLLM: Easy, Fast, and Cheap LLM Serving with PagedAttention | vLLM Blog vllm.ai. Checked 21 October 2026.
  5. Vicuna – UC Berkeley Sky Computing Lab sky.cs.berkeley.edu. Checked 21 October 2026.
  6. Judging LLM-as-a-Judge with MT-Bench and Chatbot Arena arxiv.org. Checked 21 October 2026.
  7. Starburst: A Cost-aware Scheduler for Hybrid Cloud | USENIX usenix.org. Checked 21 October 2026.
  8. Budget-aware Test-time Scaling via Discriminative Verification arxiv.org. Checked 21 October 2026.