PeopleofSuperintelligence

Researcher

Qiwei Ye

Also known as 叶启威

Qiwei Ye is a researcher at the Beijing Academy of Artificial Intelligence (BAAI) whose work covers gradient boosting, game-playing AI, protein-complex modeling and long-context language models. He co-authored the LightGBM paper at NIPS 2017 and the Suphx Mahjong paper from Microsoft Research Asia in 2020. At BAAI, he was an author of the OpenComplex code for protein and RNA complexes in 2022 and headed the Health and Bio Computing Research Center in 2023. His paper "Beyond Weisfeiler-Lehman: A Quantitative Framework for GNN Expressiveness" received an Outstanding Paper Honorable Mention at ICLR 2024, and he co-authored the Activation Beacon long-context compression module in 2024. He is a tech lead at BAAI, where he works on AGI and AI for science.

Timeline

  1. 2017

    The LightGBM paper

    He is one of eight authors of "LightGBM: A Highly Efficient Gradient Boosting Decision Tree", published at NIPS 2017 with Guolin Ke, Qi Meng, Thomas Finley, Taifeng Wang, Wei Chen, Weidong Ma and Tie-Yan Liu. The paper's Gradient-based One-Side Sampling and Exclusive Feature Bundling speed up conventional gradient boosting training by up to over 20 times with almost the same accuracy.

  2. 2020

    The Suphx Mahjong AI

    In 2020, he is the third of ten authors of the Suphx paper, work conducted at Microsoft Research Asia. Suphx is a Mahjong AI built with deep reinforcement learning, and the paper rates it above 99.99% of officially ranked human players on the Tenhou platform.

  3. 2022

    OpenComplex for biomolecules

    He is one of seven listed authors of OpenComplex in 2022, a BAAI open-source platform for protein and RNA complex models built on DeepMind's AlphaFold 2 and OpenFold. The platform adds RNA and protein-RNA complex models.

  4. 2023

    Heading BAAI health computing

    In 2023, he is head of the Health and Bio Computing Research Center at the Beijing Academy of Artificial Intelligence. He co-chairs the AI Life Sciences Forum of the 2023 BAAI Conference, which Peking University professor Xiaoliang Sunney Xie hosts.

  5. 2024

    Beyond Weisfeiler-Lehman at ICLR

    He is a co-author, with Bohang Zhang, Jingchu Gai, Yiheng Du, Di He and Liwei Wang, of "Beyond Weisfeiler-Lehman: A Quantitative Framework for GNN Expressiveness". ICLR 2024 lists the oral paper with an Outstanding Paper Honorable Mention award. It introduces homomorphism expressivity, a measure of how well graph neural networks count graphs under homomorphism.

  6. 2024

    Activation Beacon for long contexts

    In January 2024, he co-authors Activation Beacon with Peitian Zhang, Zheng Liu, Shitao Xiao, Ninglu Shao and Zhicheng Dou. The plug-in module compresses the keys and values at every layer of a transformer language model, and reaches a 2x inference speedup and an 8x reduction in KV cache memory.

  7. As of October 2026

    Tech lead at BAAI

    He is a tech lead at the Beijing Academy of Artificial Intelligence, working on AGI and AI for science. In July 2026, he is one of 24 authors of AREX, a family of recursively self-improving deep research agents.

  1. The company share of AI researchers’ top papers was 31% in 2020–2024 and 8% in 1970–1974Includes Beijing Academy of Artificial Intelligence. We estimate where the people shaping AI worked when their most-cited papers appeared: the company share was 31% in 2020–2024 and 8% in 1970–1974, and Google led all organizations in 2020–2024.
  2. Google had the most people shaping AI every year from 2014 to 2025Our estimates of which organizations employed the most people shaping AI each year from 1955 to 2025, tracing the shift from university labs to companies.
  3. The United States employs an estimated 48% of people shaping AIWe estimate where the people shaping AI work today, how universities, companies and research institutes split across countries, and how country shares have changed since the 1960s.
  4. 43% of living people shaping AI work outside their country of birthWe estimate that 43% of living people shaping AI work outside their country of birth (520 of 1,198 people with known birth and employment countries).

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 Qiwei Ye yet. Send us a free-licensed photo.

  1. LightGBM: A Highly Efficient Gradient Boosting Decision Tree proceedings.neurips.cc. Checked 22 October 2026.
  2. [2003.13590] Suphx: Mastering Mahjong with Deep Reinforcement Learning arxiv.org. Checked 22 October 2026.
  3. Suphx: Mastering Mahjong with Deep Reinforcement Learning (ar5iv) ar5iv.labs.arxiv.org. Checked 22 October 2026.
  4. GitHub - ocx-lab/OpenComplex: Trainable PyTorch framework for developing protein, RNA and complex models. github.com. Checked 22 October 2026.
  5. aeros0ul (Qiwei Ye) huggingface.co. Checked 22 October 2026.
  6. Prof. Jian Tang as the keynote speaker at 2023 BAAI Conference - BioGeometry biogeom.com. Checked 22 October 2026.
  7. ICLR 2024 Awards iclr.cc. Checked 22 October 2026.
  8. [2401.08514] Beyond Weisfeiler-Lehman: A Quantitative Framework for GNN Expressiveness arxiv.org. Checked 22 October 2026.
  9. [2401.03462] Long Context Compression with Activation Beacon arxiv.org. Checked 22 October 2026.
  10. Qiwei Ye - Google Scholar scholar.google.com. Checked 22 October 2026.
  11. [2607.21461] AREX: Towards a Recursively Self-Improving Agent for Deep Research arxiv.org. Checked 22 October 2026.