PeopleofSuperintelligence

Researcher

Yongqiang Guo

Yongqiang Guo is a researcher affiliated with DeepSeek-AI whose work spans language models and agent-training infrastructure. He coauthored the DeepSeek-V2 and DeepSeek-V3 reports in 2024; both describe mixture-of-experts models designed for efficient inference. In 2025, he coauthored DeepSeek-R1 and DeepSeek-V3.2, on reinforcement-learning reasoning and sparse attention. DeepSeek-V4 and V4.1-Flash extended that line with million-token contexts and KV-cache compression, while DeepSeek Elastic Compute described infrastructure for agentic model training. As of October 2026, his listed research affiliation is DeepSeek-AI.

Now at
DeepSeek

Timeline

  1. May 2024

    DeepSeek-V2 language model

    The May 2024 report lists Guo as a coauthor and describes DeepSeek-V2 as a 236B-parameter mixture-of-experts model. It uses 21B active parameters per token and is designed for economical training and efficient inference.

  2. December 2024

    DeepSeek-V3 technical report

    The DeepSeek-V3 report lists Guo as a coauthor in December 2024 and describes a mixture-of-experts model with 671B total parameters and 37B active per token. Its Multi-head Latent Attention and DeepSeekMoE designs support efficient inference and cost-effective training.

  3. September 2025

    The DeepSeek-R1 paper

    In September 2025, Nature publishes the DeepSeek-R1 paper with Guo among the DeepSeek-AI Team authors in Hangzhou. The paper shows how pure reinforcement learning develops reasoning patterns such as self-reflection and verification.

  4. December 2025

    Sparse attention in V3.2

    The December 2025 DeepSeek-V3.2 paper lists Guo as a coauthor and introduces DeepSeek Sparse Attention, which reduces computational complexity in long-context scenarios.

  5. April 2026

    DeepSeek-V4 model preview

    In April 2026, Guo coauthors the preview report for DeepSeek-V4-Pro and DeepSeek-V4-Flash. The models have 1.6T and 284B parameters, activate 49B and 13B per token, and support one-million-token contexts.

  6. September 2026

    KV-cache compression in V4.1-Flash

    The September 2026 report lists Guo as a coauthor of DeepSeek-V4.1-Flash, a multimodal mixture-of-experts model with 552B backbone parameters and support for contexts up to one million tokens. It combines cross-layer KV-cache reuse in Compressed Sparse Attention 2 with FP4 caching to reduce memory use.

  7. As of October 2026

    Research at DeepSeek-AI

    As of October 2026, Guo's listed research affiliation is DeepSeek-AI. In September 2026, he coauthors DeepSeek Elastic Compute (DSec), a production sandbox platform for agentic training with FnCall, container, microVM, and full-VM backends exposed through a unified SDK.

  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.

Sources and credits

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

  1. [2405.04434] DeepSeek-V2: A Strong, Economical, and Efficient Mixture-of-Experts Language Model arxiv.org. Checked 22 October 2026.
  2. [2412.19437] DeepSeek-V3 Technical Report arxiv.org. Checked 22 October 2026.
  3. DeepSeek-R1 incentivizes reasoning in LLMs through reinforcement learning | Nature nature.com. Checked 22 October 2026.
  4. [2512.02556] DeepSeek-V3.2: Pushing the Frontier of Open Large Language Models arxiv.org. Checked 22 October 2026.
  5. [2606.19348] DeepSeek-V4: Towards Highly Efficient Million-Token Context Intelligence arxiv.org. Checked 22 October 2026.
  6. [2609.19969] DeepSeek-V4.1-Flash: Pushing the Limits of KV Cache Compression arxiv.org. Checked 22 October 2026.
  7. [2609.22978] DeepSeek Elastic Compute (DSec): A Sandbox Infrastructure for Effective Agentic Training at Scale arxiv.org. Checked 22 October 2026.
  8. Paper page - DeepSeek Elastic Compute (DSec): A Sandbox Infrastructure for Effective Agentic Training at Scale | Hugging Face huggingface.co. Checked 22 October 2026.
  9. DeepSeek Elastic Compute (DSec): A Sandbox Infrastructure for Effective Agentic Training at Scale arxiv.org. Checked 22 October 2026.