PeopleofSuperintelligence

Researcher

Yixuan Tan

As of October 2026, Yixuan Tan is a researcher at DeepSeek-AI focused on model scaling and inference efficiency; their work spans language models, RDMA systems and inference. They joined DeepSeek in December 2023 as a core systems engineer and were an undergraduate in 2024. In 2024, they coauthored DeepSeek-V2, DeepSeek-V3 and RB², a high-throughput ring buffer for RDMA. In 2025, they coauthored DeepSeek-R1, which uses reinforcement learning for reasoning, and DeepSeek-V3.2, which pairs computational efficiency with reasoning and agent performance. During 2026, they coauthored DualPath, DSpark and DeepSeek-V4.1-Flash, a multimodal model that supports contexts up to one million tokens.

Now at
DeepSeek

Timeline

  1. December 2023

    Joining DeepSeek as a systems engineer

    They join DeepSeek in December 2023 as a core systems engineer. In 2024, they are an undergraduate.

  2. May 2024

    DeepSeek-V2 and RB²

    In May 2024, they coauthor DeepSeek-V2, a Mixture-of-Experts language model designed for economical training and efficient inference. At IEEE INFOCOM 2024, they coauthor RB², a high-performance RDMA-based distributed ring buffer that serves as a middle layer. It achieves 2.5× to 7.5× the throughput of state-of-the-art designs while maintaining comparable tail latency for small messages.

  3. December 2024

    The DeepSeek-V3 report

    In December 2024, they coauthor the DeepSeek-V3 report, which describes a model with 671 billion total parameters and 37 billion activated per token.

  4. 2025

    DeepSeek-R1 reasoning

    In 2025, they coauthor the DeepSeek-R1 report, which develops reasoning with pure reinforcement learning and without human-labeled reasoning trajectories. Nature publishes the peer-reviewed paper in September 2025.

  5. December 2025

    DeepSeek-V3.2 and agent performance

    In December 2025, they coauthor DeepSeek-V3.2, a model that combines high computational efficiency with reasoning and agent performance.

  6. February 2026

    DualPath for agentic inference

    In February 2026, they coauthor DualPath, which loads KV-cache data from storage into decoding engines and then transfers it to prefill engines via RDMA. The system targets storage-bandwidth limits in agentic LLM inference.

  7. July 2026

    DSpark speculative decoding

    In July 2026, they coauthor DSpark, a speculative decoding framework that combines high-throughput parallel generation with adaptive, load-aware verification.

  8. As of October 2026

    Researcher at DeepSeek-AI

    In September 2026, they coauthor DeepSeek-V4.1-Flash, a multimodal Mixture-of-Experts model with 552 billion backbone parameters and support for contexts up to one million tokens. As of October 2026, they are a researcher at DeepSeek-AI focused on model scaling and inference efficiency.

  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 Yixuan Tan yet. Send us a free-licensed photo.

  1. Deepseek's Liang Wenfeng hires graduates over "AI Wizards" efinancialcareers.com.au. Checked 21 October 2026.
  2. DeepSeek-V2: A Strong, Economical, and Efficient Mixture-of-Experts Language Model arxiv.org. Checked 21 October 2026.
  3. IEEE INFOCOM 2024 infocom.info. Checked 21 October 2026.
  4. DeepSeek-V3 Technical Report arxiv.org. Checked 21 October 2026.
  5. DeepSeek-R1: Incentivizing Reasoning Capability in LLMs via Reinforcement Learning arxiv.org. Checked 21 October 2026.
  6. DeepSeek-R1 incentivizes reasoning in LLMs through reinforcement learning nature.com. Checked 21 October 2026.
  7. DeepSeek-V3.2: Pushing the Frontier of Open Large Language Models arxiv.org. Checked 21 October 2026.
  8. DualPath: Breaking the Storage Bandwidth Bottleneck in Agentic LLM Inference arxiv.org. Checked 21 October 2026.
  9. DSpark: Confidence-Scheduled Speculative Decoding with Semi-Autoregressive Generation arxiv.org. Checked 21 October 2026.
  10. DeepSeek-V4.1-Flash: Pushing the Limits of KV Cache Compression arxiv.org. Checked 21 October 2026.
  11. Deepseek researchers | alphaXiv alphaxiv.org. Checked 21 October 2026.