PeopleofSuperintelligence

Researcher

Zihui Gu

Zihui Gu is a researcher whose AI work spans table-based fact verification, database queries, and instruction-following evaluation. He coauthored PASTA in 2022 and work on few-shot Text-to-SQL and multimodal data lakes in 2023. In 2024, he coauthored DINGO and the DeepSeek-V2 and DeepSeek-V3 papers; his 2025 DeepSeek-R1 paper showed how pure reinforcement learning can improve language-model reasoning without human-labeled reasoning trajectories. He completed an MS at Renmin University of China in 2024, coauthored a 2026 DeepSeek-V4.1-Flash paper on KV-cache compression, and works at DeepSeek.

Now at
DeepSeek
Badges

Timeline

  1. 2022

    PASTA for table verification

    In 2022, he coauthors PASTA, which pre-trains DeBERTaV3 on synthesized sentence–table cloze questions for table-based fact verification.

  2. 2023

    Few-shot Text-to-SQL prompts

    In 2023, he coauthors a few-shot Text-to-SQL paper that uses structure and content prompt learning.

  3. 2023

    Symphony for data lakes

    In 2023, he coauthors “Symphony: Towards Natural Language Query Answering over Multi-modal Data Lakes.”

  4. 2024

    DINGO instruction evaluation

    In 2024, he coauthors DINGO, a fine-grained and diverse dataset for evaluating instruction-following on real-world user requests.

  5. May 2024

    The DeepSeek-V2 model

    In May 2024, he coauthors DeepSeek-V2, a mixture-of-experts model with 236B total parameters and 21B activated for each token.

  6. December 2024

    DeepSeek-V3 technical report

    In December 2024, he coauthors the DeepSeek-V3 Technical Report, which describes a mixture-of-experts model with 671B total and 37B activated parameters per token.

  7. 2025

    DeepSeek-R1 reasoning

    In 2025, he coauthors DeepSeek-R1, which incentivizes reasoning in language models through pure reinforcement learning without human-labeled reasoning trajectories.

  8. As of October 2026

    DeepSeek and V4.1-Flash

    He completes an MS at Renmin University of China in 2024. In September 2026, he coauthors DeepSeek-V4.1-Flash, a paper on KV-cache compression. He works at DeepSeek.

Career

  1. NowDeepSeek

Education

  1. 2024Renmin University of ChinaMS

From the biography.

  1. The median h-index is 48 at universities and 14 at frontier AI labsIncludes Zihui Gu. 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 Zihui Gu. 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.

Sources and credits

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

  1. PASTA: Table-Operations Aware Fact Verification via Sentence-Table Cloze Pre-training | ACL Anthology aclanthology.org. Checked 21 October 2026.
  2. DataLab | Publications ruc-datalab.com. Checked 21 October 2026.
  3. DINGO: Towards Diverse and Fine-Grained Instruction-Following Evaluation | Proceedings of the AAAI Conference on Artificial Intelligence ojs.aaai.org. Checked 21 October 2026.
  4. DeepSeek-V2: A Strong, Economical, and Efficient Mixture-of-Experts Language Model arxiv.org. Checked 21 October 2026.
  5. DeepSeek-V3 Technical Report arxiv.org. Checked 21 October 2026.
  6. DeepSeek-R1: Incentivizing Reasoning Capability in LLMs via Reinforcement Learning arxiv.org. Checked 21 October 2026.
  7. DataLab | People ruc-datalab.com. Checked 21 October 2026.
  8. DeepSeek-V4.1-Flash: Pushing the Limits of KV Cache Compression arxiv.org. Checked 21 October 2026.