PeopleofSuperintelligence

Researcher

Zeyu Cui

Zeyu Cui is a member of the Alibaba Qwen Team, where he works on code-focused language models and coding agents. He first worked on graph neural networks at the Institute of Automation of the Chinese Academy of Sciences, first-authoring the 2019 outfit-compatibility paper Dressing as a Whole, and he co-authored the text-classification method TextING in 2020. In 2022, he first-authored M6-Rec at Alibaba's DAMO Academy, and in 2023 he was among the authors of the Qwen Technical Report. The Qwen2.5-Coder, Qwen2.5, Qwen3 and Qwen3-Coder-Next reports followed with his name on the author list, and in 2025 he co-authored the CodeElo benchmark for competition-level code generation. In June 2026, he was a core contributor to The Verification Horizon, a Qwen Team paper on reward design for coding agents.

Now at
Alibaba

Timeline

  1. 2019

    Outfit compatibility with graph networks

    At the Institute of Automation, Chinese Academy of Sciences, he is first author of Dressing as a Whole, written with Zekun Li, Shu Wu, Xiaoyu Zhang and Liang Wang. The paper, accepted at the 2019 World Wide Web Conference, learns outfit compatibility with node-wise graph neural networks.

  2. 2020

    TextING for text classification

    With Yufeng Zhang, Xueli Yu, Shu Wu, Zhongzhen Wen and Liang Wang, he co-authors TextING, a graph neural network method for inductive text classification that builds an individual graph for each document. The paper appears at ACL 2020.

  3. 2022

    Recommenders on the M6 model

    At Alibaba's DAMO Academy, he is first author of M6-Rec, written in 2022 with Jianxin Ma, Chang Zhou, Jingren Zhou and Hongxia Yang. The paper builds a unified foundation model for industrial recommender systems on M6, a large-scale pretrained language model similar to GPT-3 and T5.

  4. 2023

    The first Qwen report

    In 2023, he is among the authors of the Qwen Technical Report, which introduces Qwen as the first installment of a large language model series from Alibaba Cloud. The series encompasses distinct models with varying parameter counts.

  5. 2024

    Qwen2.5-Coder and Qwen2.5

    In September 2024, he is the third author of the Qwen2.5-Coder Technical Report, which continues the pretraining of Qwen2.5 on over 5.5 trillion tokens to build code models. In December 2024, he is an author of the Qwen2.5 Technical Report.

  6. 2025

    CodeElo and the Qwen3 report

    In January 2025, he co-authors CodeElo, a benchmark that rates language models on competition-level code generation with human-comparable Elo ratings and covers 30 open-source and 3 proprietary models. In May 2025, he is an author of the Qwen3 Technical Report on the latest Qwen model family.

  7. February 2026

    Qwen3-Coder-Next report

    In February 2026, he is among the authors of the Qwen3-Coder-Next Technical Report, which presents an open-weight coding-agent model with 80 billion parameters that activates 3 billion during inference. The Qwen3-Coder repository on GitHub, the code version of Qwen3 developed by the Qwen team, cites the report.

  8. June 2026

    Reward design for coding agents

    In June 2026, he is a core contributor to The Verification Horizon, a paper from the Alibaba Qwen Team on verification signals for coding agents. It studies four reward constructions: test verifiers, rubric verifiers, user feedback and automated agent verifiers.

Career

  1. Since 2022AlibabaMember
  2. From 2019Chinese Academy of Sciences

From the biography.

  1. Google alumni have founded or led 133 other organizations, more than any other AI labIncludes Alibaba. 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.
  2. Universities (31%) and startups and other companies (25%) employ the largest shares of 2020–2026 entrantsIncludes Alibaba. Among 819 people whose first recorded role started in 2020–2026 and who have a known current organization, 60% work in commercial sectors (25% at startups and other companies, 23% at frontier labs and 13% at big tech, with 9 in two), and 31% work at universities, compared with 66% for earlier entrants.
  3. The top 10 organizations employed 22% of people shaping AI in 2025: 51% of company researchers and 21% of university researchersIncludes Alibaba. We estimate the share of people shaping AI who worked at the top 10 organizations each year from 1955 to 2025, comparing concentration across companies, universities, and institutes and labs.
  4. 16% of AI career moves in the 2020s went from a university to a companyIncludes Alibaba. 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.

Colleagues

People named in this biography, and people with dated positions at the same organization in the same years.

Sources and credits

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

  1. [1902.08009] Dressing as a Whole: Outfit Compatibility Learning Based on Node-wise Graph Neural Networks arxiv.org. Checked 21 October 2026.
  2. Dressing as a Whole: Outfit Compatibility Learning Based on Node-wise Graph Neural Networks (HTML version) ar5iv.labs.arxiv.org. Checked 21 October 2026.
  3. [2004.13826] Every Document Owns Its Structure: Inductive Text Classification via Graph Neural Networks arxiv.org. Checked 21 October 2026.
  4. Every Document Owns Its Structure: Inductive Text Classification via Graph Neural Networks - ACL Anthology aclanthology.org. Checked 21 October 2026.
  5. [2205.08084] M6-Rec: Generative Pretrained Language Models are Open-Ended Recommender Systems arxiv.org. Checked 21 October 2026.
  6. M6-Rec: Generative Pretrained Language Models are Open-Ended Recommender Systems (HTML version) ar5iv.labs.arxiv.org. Checked 21 October 2026.
  7. [2309.16609] Qwen Technical Report arxiv.org. Checked 21 October 2026.
  8. GitHub - QwenLM/Qwen: The official repo of Qwen (通义千问) chat & pretrained large language model proposed by Alibaba Cloud github.com. Checked 21 October 2026.
  9. [2409.12186] Qwen2.5-Coder Technical Report arxiv.org. Checked 21 October 2026.
  10. [2412.15115] Qwen2.5 Technical Report arxiv.org. Checked 21 October 2026.
  11. [2501.01257] CodeElo: Benchmarking Competition-level Code Generation of LLMs with Human-comparable Elo Ratings arxiv.org. Checked 21 October 2026.
  12. [2505.09388] Qwen3 Technical Report arxiv.org. Checked 21 October 2026.
  13. [2603.00729] Qwen3-Coder-Next Technical Report arxiv.org. Checked 21 October 2026.
  14. GitHub - QwenLM/Qwen3-Coder: Qwen3-Coder is the code version of Qwen3, the large language model series developed by Qwen team github.com. Checked 21 October 2026.
  15. The Verification Horizon: No Silver Bullet for Coding Agent Rewards (HTML version) arxiv.org. Checked 21 October 2026.
  16. [2606.26300] The Verification Horizon: No Silver Bullet for Coding Agent Rewards arxiv.org. Checked 21 October 2026.