PeopleofSuperintelligence

Lab leader

Bowen Yu

Also known as 郁博文

Bowen Yu is an AI researcher whose work spans relation extraction and large language model training. He published early work on joint entity-and-relation extraction and earned a PhD from the Institute of Information Engineering, Chinese Academy of Sciences, in 2022. From 2022 to March 2026, he served as an Algorithm Expert at Alibaba's Qwen team and led Qwen post-training research. Across 2023–2025, he coauthored API-Bank, Preference Ranking Optimization, the Qwen3 technical report, and GSPO, a sequence-level reinforcement-learning algorithm. Since March 2026, he leads post-training for visual models and multimodal interaction on ByteDance Seed.

Now at
ByteDance
Works
109
Citations
1,380
h-index
19
Badges

Publication numbers from OpenAlex, as of 19 October 2026.

Timeline

  1. 2019–2020

    Decomposing relation extraction

    In 2019, he and six coauthors submit a paper that splits joint entity-and-relation extraction into two linked tasks; ECAI accepts it in 2020.

  2. 2022

    A doctorate in information engineering

    In 2022, he earns a PhD from the Institute of Information Engineering, Chinese Academy of Sciences, under Tingwen Liu and Bin Wang.

  3. 2022–March 2026

    An algorithm expert at Qwen

    From 2022 to March 2026, he serves as an Algorithm Expert at Alibaba's Qwen team and leads Qwen post-training.

  4. 2023

    A benchmark for tool use

    In 2023, he coauthors API-Bank, a benchmark with 73 API tools that evaluates how language models plan, retrieve, and call external APIs.

  5. 2024

    Preference ranking for alignment

    In 2024, he coauthors Preference Ranking Optimization (PRO), an SFT method that fine-tunes language models against ranked human preferences.

  6. 2025

    The Qwen3 technical report

    In 2025, he coauthors the Qwen3 technical report, which presents dense and mixture-of-experts models with both thinking and non-thinking modes.

  7. July 2025

    Sequence-level reinforcement learning

    In July 2025, he coauthors Group Sequence Policy Optimization (GSPO), an RL algorithm that defines training updates at the sequence level for language models.

  8. Since March 2026

    Leading post-training at Seed

    Since March 2026, he leads post-training for visual models and multimodal interaction on ByteDance's Seed team.

Career

  1. Since 2026ByteDancePost-training lead
  2. 2023–2025Nanjing University
  3. 2022–2026AlibabaAlgorithm Expert
  4. 2022–2025University of Michigan
  5. 2022–2026University of Michigan–Dearborn
  6. 2021–2023University of Chinese Academy of Sciences
  7. 2020–2026Wuhan University

Education

  1. 2022Chinese Academy of SciencesPhD

From the biography and OpenAlex affiliations (the years of papers that list the organization), so some years are approximate.

Most-cited works

  1. Qwen Technical Report2023 · 331 citationsWith Binyuan Hui, Fei Huang and Rui Men
  2. API-Bank: A Comprehensive Benchmark for Tool-Augmented LLMs2023 · Conference on Empirical Methods in Natural Language Processing (EMNLP) · 120 citationsWith Fei Huang
  3. Intergenic CircRNA Circ_0007379 Inhibits Colorectal Cancer Progression by Modulating miR-320a Biogenesis in a KSRP-Dependent Manner2023 · International Journal of Biological Sciences · 77 citations
  4. Preference Ranking Optimization for Human Alignment2024 · Proceedings of the AAAI Conference on Artificial Intelligence · 77 citationsWith Fei Huang
  5. Pre-Trained Model-Based Automated Software Vulnerability Repair: How Far are We?2023 · IEEE Transactions on Dependable and Secure Computing · 59 citations

From OpenAlex, as of 19 October 2026.

  1. Google alumni founded or led 81 other organizations, the most among the labs we compareIncludes ByteDance. We trace 194 organizations founded or led by AI lab alumni, and 202 career moves between the labs.
  2. The company share of AI researchers’ top papers rose from 2% to 32%Includes ByteDance. We estimate where leading AI researchers worked when their most-cited papers appeared. The company share rose from 2% in 1975–1979 to 32% in 2020–2024.
  3. Stanford University trained or employed 26 of the 108 AI major-prize winners, more than any other universityIncludes Wuhan University. Stanford University trained or employed 26 of the 108 AI major-prize winners, more than any other university. We compare doctorate and employment links for major-prize and best-paper winners.
  4. Google has the highest combined h-index among AI organizationsIncludes Alibaba. Our estimates of combined and median h-index at AI organizations, using current affiliations and OpenAlex records.

Colleagues

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

Sources and credits

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

  1. Joint Extraction of Entities and Relations Based on a Novel Decomposition Strategy arxiv.org. Checked 18 October 2026.
  2. Bowen Yu @ Qwen Team, Alibaba Group yubowen-ph.github.io. Checked 18 October 2026.
  3. Bowen Yu | alphaXiv alphaxiv.org. Checked 18 October 2026.
  4. API-Bank: A Comprehensive Benchmark for Tool-Augmented LLMs arxiv.org. Checked 18 October 2026.
  5. Preference Ranking Optimization for Human Alignment ojs.aaai.org. Checked 18 October 2026.
  6. Qwen3 Technical Report arxiv.org. Checked 18 October 2026.
  7. GSPO: Towards Scalable Reinforcement Learning for Language Models qwenlm.github.io. Checked 18 October 2026.
  8. ByteDance hires former Qwen lead kr-asia.com. Checked 18 October 2026.