PeopleofSuperintelligence

Builder

Zhean Xu

Also known as CyanicX

Zhean Xu is a Core Engineer at DeepSeek AI in Hangzhou who works on machine learning systems and algorithms. A paper on faster knapsack algorithms, written with Qizheng He, appeared in August 2023, and Xu joined DeepSeek in October 2023. DeepSeek-V3 in December 2024 and DeepSeek-R1 in January 2025, which Nature published in September 2025, both have Xu among their authors. DeepEP, an expert-parallel communication library published on GitHub in 2025, has Xu among its authors, as do DeepSeek-V3.2 and mHC from December 2025. Xu is also an author of DeepSeek-V4.1-Flash, posted in September 2026.

Now at
DeepSeek
Links
Google Scholar, X and LinkedIn

Timeline

  1. August 2023

    Faster knapsack algorithms

    Simple and Faster Algorithms for Knapsack, posted on arXiv in August 2023 by Qizheng He and Xu, presents new pseudo-polynomial time algorithms for the knapsack problem.

  2. October 2023

    Joining DeepSeek AI

    DeepSeek AI in Hangzhou is where Xu starts work as a Core Engineer in October 2023.

  3. December 2024

    The DeepSeek-V3 technical report

    The DeepSeek-V3 technical report appears on arXiv in December 2024 with Xu among its authors. DeepSeek-V3 is a Mixture-of-Experts language model with 671B total parameters, 37B of them activated for each token.

  4. January 2025–September 2025

    DeepSeek-R1 on arXiv and Nature

    DeepSeek-R1 appears on arXiv in January 2025 with Xu among its authors. The paper shows that the reasoning abilities of LLMs can be incentivized through pure reinforcement learning, and a version in Nature follows in September 2025.

  5. 2025

    The DeepEP library

    In 2025, DeepSeek publishes DeepEP, an efficient expert-parallel communication library, on GitHub with Xu among its nine listed authors. Systems optimization work by Xu also includes DeepGEMM, which implements FP8 GEMM kernels with fine-grained scaling.

  6. December 2025

    DeepSeek-V3.2 and mHC

    DeepSeek-V3.2, a model that combines DeepSeek Sparse Attention with a scalable reinforcement learning framework, appears in December 2025 with Xu among its authors. The mHC paper from December 2025 also has Xu as an author and proposes Manifold-Constrained Hyper-Connections, which restore the identity mapping property of the residual connection.

  7. January 2026

    A quantum data structure

    Quantum Data Structure for Range Minimum Query, posted in January 2026 by Qisheng Wang, Xu and Zhicheng Zhang, proposes a quantum data structure that supports range minimum queries and range updates with an optimal time complexity.

  8. As of October 2026

    Core Engineer at DeepSeek

    Xu is a Core Engineer at DeepSeek AI in Hangzhou, working on machine learning systems and algorithms. DSpark, a speculative decoding framework posted in July 2026, and DeepSeek-V4.1-Flash, a multimodal Mixture-of-Experts model with 552B backbone parameters posted in September 2026, both have Xu among their authors.

  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 Zhean Xu yet. Send us a free-licensed photo.

  1. [2308.11307] Simple and Faster Algorithms for Knapsack arxiv.org. Checked 20 October 2026.
  2. Zhean Xu | alphaXiv alphaxiv.org. Checked 20 October 2026.
  3. [2412.19437] DeepSeek-V3 Technical Report arxiv.org. Checked 20 October 2026.
  4. [2501.12948] DeepSeek-R1: Incentivizing Reasoning Capability in LLMs via Reinforcement Learning arxiv.org. Checked 20 October 2026.
  5. DeepSeek-R1 incentivizes reasoning in LLMs through reinforcement learning | Nature nature.com. Checked 20 October 2026.
  6. GitHub - deepseek-ai/DeepEP: DeepEP: an efficient expert-parallel communication library github.com. Checked 20 October 2026.
  7. [2512.02556] DeepSeek-V3.2: Pushing the Frontier of Open Large Language Models arxiv.org. Checked 20 October 2026.
  8. [2512.24880] mHC: Manifold-Constrained Hyper-Connections arxiv.org. Checked 20 October 2026.
  9. [2601.13195] Quantum Data Structure for Range Minimum Query arxiv.org. Checked 20 October 2026.
  10. [2607.05147] DSpark: Confidence-Scheduled Speculative Decoding with Semi-Autoregressive Generation arxiv.org. Checked 20 October 2026.
  11. [2609.19969] DeepSeek-V4.1-Flash: Pushing the Limits of KV Cache Compression arxiv.org. Checked 20 October 2026.
  12. Zhean Xu - Google Scholar scholar.google.com. Checked 20 October 2026.