PeopleofSuperintelligence

Researcher

Wanjia Zhao

Also known as 赵万佳

Wanjia Zhao is a computer science researcher whose work spans language-model reasoning, formal theorem proving, and AI for science. They earned an honors mathematics degree at Zhejiang University, researched physics-informed learning at UCLA, and began a computer science PhD at Stanford in 2024. At DeepSeek, they coauthored DeepSeek-V3, DeepSeek-R1, and DeepSeek-Prover-V2; later projects include multi-agent reasoning, geometric diffusion, and protein language models. Since January 2026, they work as a student researcher at Google DeepMind and continue work on physics reasoning with PRISM-Physics.

Badges

Timeline

  1. 2020–2024

    Mathematics at Zhejiang University

    From September 2020 to June 2024, they study mathematics and applied mathematics at Zhejiang University and earn an honors BS. Professor Fei Wu advises them.

  2. June 2023–October 2023

    Research at UCLA

    From June to October 2023, they are a visiting research student at UCLA's Scalable Analytics Institute (ScAi), where they work on physics-informed machine learning under Yizhou Sun and Wei Wang.

  3. 2023–2024

    A physics-informed NeurIPS paper

    They coauthor Physics-Informed Regularization for Domain-Agnostic Dynamical System Modeling, which wins Best Paper at the NeurIPS 2023 Deep Learning and Differential Equations workshop. The paper appears in the NeurIPS 2024 proceedings.

  4. February 2024–December 2024

    Microsoft and DeepSeek internships

    From February to June 2024, they work in Microsoft Research Lab–Asia's Machine Learning Group on LLMs for decision making. From June to September 2024, they research mathematical reasoning as a DeepSeek AI intern. DeepSeek-V3's December 2024 technical report lists them among its research and engineering contributors.

  5. September 2024

    A Stanford PhD and award

    They begin a computer science PhD at Stanford in September 2024, with affiliations at the Stanford NLP Group and Stanford AI Lab and advisors James Zou and Ludwig Schmidt. Stanford lists them among Zou's doctoral students, and they receive the NeurIPS 2024 Scholar Award.

  6. January 2025–April 2025

    DeepSeek reasoning and theorem proving

    In January 2025, they coauthor DeepSeek-R1, which describes reinforcement-learning training for reasoning models. In April 2025, they coauthor DeepSeek-Prover-V2, an open-source Lean 4 model for formal theorem proving through subgoal decomposition.

  7. June 2025–November 2025

    Microsoft research and 2025 papers

    From June to September 2025, they are an AI Frontiers research intern at Microsoft Research, working on reinforcement learning and tool use for LLM reasoning. Their NeurIPS 2025 papers include SiriuS, a self-improving multi-agent reasoning system, and GeoAda, an equivariant adapter method for geometric diffusion models. In November 2025, they coauthor Protein Large Language Models: A Comprehensive Survey, which covers architectures, training data, evaluation metrics, and applications.

  8. Since January 2026

    Google DeepMind research

    Since January 2026, they work as a student researcher at Google DeepMind while continuing their Stanford computer science PhD. Their ICLR 2026 paper PRISM-Physics develops a process-level benchmark that represents physics solutions as directed acyclic graphs of formulas.

Career

  1. Since 2026Google DeepMindStudent Researcher
  2. 2025Microsoft ResearchAI Frontiers Research Intern
  3. 2024Microsoft Research AsiaResearch Intern
  4. 2024DeepSeekAI Intern
  5. 2023University of California, Los AngelesVisiting Research Student

Education

  1. 2024Zhejiang UniversityHonors BS
  2. From 2024Stanford UniversityPhD in computer science

From the biography.

  1. The median h-index is 48 at universities and 14 at frontier AI labsIncludes Wanjia Zhao. 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.
  2. 16% of AI career moves in the 2020s went from a university to a companyIncludes Wanjia Zhao. We estimate that 16% of AI career moves in the 2020s went from a university to a company; 29% of moves starting at a university went to a company, and 76% ended the university role.
  3. Google had the most people shaping AI every year from 2014 to 2025Includes Google. Our estimates of which organizations employed the most people shaping AI each year from 1955 to 2025, tracing the shift from university labs to companies.
  4. Google has the highest combined h-index among AI organizationsIncludes Google. We estimate that Google holds the highest combined h-index at 8,016, while universities account for 73% of the deduplicated total across AI organizations.

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 Wanjia Zhao yet. Send us a free-licensed photo.

  1. Wanjia Zhao 赵万佳 wanjiazhao1203.github.io. Checked 20 October 2026.
  2. Physics-Informed Regularization for Domain-Agnostic Dynamical System Modeling proceedings.neurips.cc. Checked 20 October 2026.
  3. DeepSeek-V3 Technical Report arxiv.org. Checked 20 October 2026.
  4. James Zou's Profile | Stanford Profiles profiles.stanford.edu. Checked 20 October 2026.
  5. DeepSeek-R1: Incentivizing Reasoning Capability in LLMs via Reinforcement Learning arxiv.org. Checked 20 October 2026.
  6. DeepSeek-Prover-V2: Advancing Formal Mathematical Reasoning via Reinforcement Learning for Subgoal Decomposition arxiv.org. Checked 20 October 2026.
  7. Protein Large Language Models: A Comprehensive Survey - ACL Anthology aclanthology.org. Checked 20 October 2026.
  8. PRISM-Physics: Causal DAG-Based Process Evaluation for Physics Reasoning arxiv.org. Checked 20 October 2026.