PeopleofSuperintelligence

Researcher

Runxin Xu

Also known as 许润昕

Runxin Xu is a researcher at DeepSeek who works on large language models. They earned a bachelor's degree at Shanghai Jiao Tong University in 2020 and a master's degree in 2023 at Peking University's Institute of Computational Linguistics, advised by Baobao Chang and Zhifang Sui. After joining DeepSeek in August 2023, they coauthored DeepSeekMoE, DeepSeekMath, DeepSeek-V2 and DeepSeek-V3 in 2024, and the DeepSeek-V4 report in 2026. They were a core contributor to DeepSeek-R1, whose paper Nature published in September 2025.

Now at
DeepSeek
Works
56
Citations
2,986
h-index
14
Badges

Publication numbers from OpenAlex, as of 22 October 2026.

Timeline

  1. 2016–2020

    Engineering degree at Shanghai Jiao Tong

    Between 2016 and 2020, they study for a B.Eng. in the School of Cyber Science and Engineering at Shanghai Jiao Tong University.

  2. 2019–2021

    Internships at ByteDance and Alibaba

    They are a research intern at ByteDance AI Lab from November 2019 to January 2021, working on information extraction and machine translation. In 2021, they intern at Alibaba DAMO Academy on effective and efficient language models.

  3. 2020–2023

    Master's at Peking University

    In 2020, they begin an M.S. in the School of EECS at Peking University, working in the Institute of Computational Linguistics under Baobao Chang and Zhifang Sui. They finish the degree in 2023.

  4. August 2023

    Joining DeepSeek as a researcher

    In August 2023, they join DeepSeek as a researcher working on large language models.

  5. 2024

    DeepSeekMoE and DeepSeekMath

    In February 2024, they coauthor DeepSeekMath, whose paper introduces Group Relative Policy Optimization (GRPO) and a data selection pipeline for publicly available web data. They are also an author of DeepSeekMoE, published at ACL 2024.

  6. May 2024–December 2024

    DeepSeek-V2 and DeepSeek-V3

    In May 2024, they are among the authors of DeepSeek-V2, a Mixture-of-Experts model with 236B total parameters and 21B activated for each token. In December 2024, they are among the authors of DeepSeek-V3, which has 671B total parameters with 37B activated for each token.

  7. September 2025

    DeepSeek-R1 in Nature

    Nature publishes the DeepSeek-R1 paper in September 2025, and Runxin Xu is listed among its core contributors. The paper shows that the reasoning abilities of LLMs can be incentivized through pure reinforcement learning (RL).

  8. As of October 2026

    Researcher at DeepSeek

    They are a researcher at DeepSeek, working on large language models. In 2026, Runxin Xu is among the authors of the DeepSeek-V4 report, which presents two Mixture-of-Experts models, V4-Pro with 1.6T parameters and V4-Flash with 284B parameters, both supporting a context length of one million tokens.

Career

  1. Since 2023DeepSeekResearcher
  2. 2025–2026Shanghai Jiao Tong University
  3. 2022–2023Imperial College London
  4. From 2021AlibabaIntern
  5. 2019–2021ByteDanceResearch intern

Education

  1. 2023Peking UniversityM.S.
  2. 2020Shanghai Jiao Tong UniversityB.Eng.

From the biography and OpenAlex affiliations (the years of papers that list the organization), so some years are approximate.

Most-cited works

  1. DeepSeek-R1 incentivizes reasoning in LLMs through reinforcement learning2025 · Nature · 1,118 citationsWith Bingxuan Wang, Chengqi Deng, Damai Dai, Daya Guo, Erhang Li and Fangyun Lin
  2. Can Open Large Language Models Catch Vulnerabilities?2025 · 539 citationsWith Bingxuan Wang, Chengqi Deng, Damai Dai, Erhang Li, Fangyun Lin and Haocheng Wang
  3. DeepSeek-V3 Technical Report2024 · 292 citationsWith Bingxuan Wang, Chengqi Deng, Damai Dai, Daya Guo, Dejian Yang and Erhang Li
  4. Double Graph Based Reasoning for Document-level Relation Extraction2020 · Conference on Empirical Methods in Natural Language Processing (EMNLP) · 261 citations
  5. Raise a Child in Large Language Model: Towards Effective and Generalizable Fine-tuning2021 · Conference on Empirical Methods in Natural Language Processing (EMNLP) · 137 citationsWith Fei Huang

From OpenAlex, as of 21 October 2026.

  1. The median h-index is 48 at universities and 14 at frontier AI labsIncludes Runxin Xu. 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 Runxin Xu. 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. Companies supply the most prior-work hires at 7 of the 9 frontier AI labsIncludes DeepSeek. 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.
  4. 5,000 frontier users would be 38.2% of the core labs' 13,100 reported staffIncludes DeepSeek. 5,000 frontier users would be 38.2% of the core labs' 13,100 reported staff. We test the estimate against 3,913 named model contributors and dated all-staff records across Google DeepMind, OpenAI and Anthropic, plus six additional frontier-grade labs.

Colleagues

People with dated positions at the same organization in the same years.

Sources and credits

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

  1. Runxin Xu (许润昕) runxinxu.github.io. Checked 20 October 2026.
  2. Runxin Xu | alphaXiv alphaxiv.org. Checked 20 October 2026.
  3. [2402.03300] DeepSeekMath: Pushing the Limits of Mathematical Reasoning in Open Language Models arxiv.org. Checked 20 October 2026.
  4. [2405.04434] DeepSeek-V2: A Strong, Economical, and Efficient Mixture-of-Experts Language Model arxiv.org. Checked 20 October 2026.
  5. [2412.19437] DeepSeek-V3 Technical Report arxiv.org. Checked 20 October 2026.
  6. DeepSeek-R1 incentivizes reasoning in LLMs through reinforcement learning | Nature nature.com. Checked 20 October 2026.
  7. [2606.19348] DeepSeek-V4: Towards Highly Efficient Million-Token Context Intelligence arxiv.org. Checked 20 October 2026.