PeopleofSuperintelligence

Lab leader

Yishi Piao

Yishi Piao is a DeepSeek author whose coauthored work spans language, code, vision-language and reasoning models. In January 2024, he coauthored DeepSeek LLM with 7B and 67B configurations; in May 2024, he coauthored DeepSeek-V2, a 236B mixture-of-experts model with 21B active parameters per token. He also coauthored DeepSeek-Coder-V2, DeepSeek-V3, DeepSeek-VL2, DeepSeek-R1, DeepSeek-V3.2, DeepSeek-V4 and DeepSeek-V4.1-Flash across 2024–2026. Since 2025, he is Head of Data at DeepSeek.

Now at
DeepSeek

Timeline

  1. January 2024–May 2024

    DeepSeek LLM and DeepSeek-V2

    In January 2024, he coauthors DeepSeek LLM, a project to advance open-source language models with 7B and 67B configurations. In May 2024, he coauthors DeepSeek-V2, a 236B mixture-of-experts model with 21B active parameters per token that uses Multi-head Latent Attention and DeepSeekMoE.

  2. June 2024

    DeepSeek-Coder-V2 for code

    In June 2024, he coauthors DeepSeek-Coder-V2, an open-source mixture-of-experts code model that achieves performance comparable to GPT4-Turbo on code-specific tasks.

  3. December 2024

    DeepSeek-VL2 for vision

    In December 2024, he coauthors DeepSeek-VL2, a vision-language model that processes high-resolution images and shows capabilities in visual question answering, optical character recognition and document understanding.

  4. December 2024

    DeepSeek-V3 Technical Report

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

  5. January 2025

    Reasoning with DeepSeek-R1

    In January 2025, he coauthors DeepSeek-R1, showing that pure reinforcement learning can incentivize reasoning in language models without human-labeled reasoning trajectories.

  6. December 2025

    DeepSeek-V3.2 sparse attention

    In December 2025, he coauthors DeepSeek-V3.2, a model for computational efficiency, reasoning and agent performance. DeepSeek Sparse Attention reduces computation in long-context scenarios.

  7. 2026

    DeepSeek-V4 and V4.1-Flash

    In 2026, he coauthors DeepSeek-V4, whose Pro version supports a one-million-token context. In September 2026, he coauthors DeepSeek-V4.1-Flash, a 552B multimodal mixture-of-experts model with context support up to one million tokens.

  8. Since 2025

    Head of Data at DeepSeek

    At GOSIM Hangzhou in September 2025, he is Head of Data at DeepSeek and joins Session 5 on building open and compatible digital public products.

  1. The median h-index is 48 at universities and 14 at frontier AI labsIncludes DeepSeek. 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. 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.
  3. 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.
  4. Google alumni have founded or led 133 other organizations, more than any other AI labIncludes DeepSeek. We estimate that alumni of 17 major AI labs went on to found or lead 350 other organizations, including 226 they founded, and we trace 472 career moves directly between 23 labs since 2010.

Colleagues

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

Sources and credits

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

  1. DeepSeek LLM: Scaling Open-Source Language Models with Longtermism arxiv.org. Checked 22 October 2026.
  2. DeepSeek-V2: A Strong, Economical, and Efficient Mixture-of-Experts Language Model arxiv.org. Checked 22 October 2026.
  3. DeepSeek-Coder-V2: Breaking the Barrier of Closed-Source Models in Code Intelligence arxiv.org. Checked 22 October 2026.
  4. DeepSeek-VL2: Mixture-of-Experts Vision-Language Models for Advanced Multimodal Understanding arxiv.org. Checked 22 October 2026.
  5. DeepSeek-V3 Technical Report arxiv.org. Checked 22 October 2026.
  6. DeepSeek-V3 Technical Report | European Laboratory for Learning and Intelligent Systems ellis.eu. Checked 22 October 2026.
  7. DeepSeek-R1: Incentivizing Reasoning Capability in LLMs via Reinforcement Learning arxiv.org. Checked 22 October 2026.
  8. DeepSeek-V3.2: Pushing the Frontier of Open Large Language Models arxiv.org. Checked 22 October 2026.
  9. DeepSeek-V4: Towards Highly Efficient Million-Token Context Intelligence arxiv.org. Checked 22 October 2026.
  10. DeepSeek-V4.1-Flash: Pushing the Limits of KV Cache Compression arxiv.org. Checked 22 October 2026.
  11. Piao Yishi – GOSIM Hangzhou 2025 hangzhou2025.gosim.org. Checked 22 October 2026.