PeopleofSuperintelligence

Researcher

Zhenru Zhang

Also known as Zhen-Ru Zhang

Zhenru Zhang is an AI researcher whose work spans semi-supervised learning, language-model tuning, mathematical reasoning and agents. They coauthored early work on partial-label learning and demonstration tuning, then Alibaba's Qwen technical reports through Qwen3. ProcessBench and Qwen2.5-Math-PRM brought their research into mathematical reasoning, while DeepMiner and Qwen-CUA carried it into search and computer use. In September 2026, they led Terminal-Universe, a framework for turning agent trajectories into reusable terminal environments. As of October 2026, they are a researcher in Alibaba's Qwen Team.

Timeline

  1. 2021–2023

    Partial-label learning and Qwen

    In 2021, they coauthor a paper on partial-label assignment for semi-supervised learning. In 2022, they coauthor Contrastive Demonstration Tuning, a prompt-tuning method that does not use demonstration sampling. In September 2023, they coauthor the Qwen Technical Report.

  2. July 2024–December 2024

    Qwen2 and Qwen2.5 reports

    They coauthor the Qwen2 Technical Report in July 2024, the Qwen2.5-Math Technical Report in September 2024, and the Qwen2.5 Technical Report in December 2024.

  3. December 2024

    ProcessBench for math errors

    In December 2024, they coauthor ProcessBench, a 3,400-case benchmark with human-annotated solutions that tests whether models find the earliest erroneous step.

  4. January 2025

    Process reward models

    In January 2025, they coauthor a study that introduces process reward models for mathematical reasoning at 7B and 72B scales.

  5. May 2025–October 2025

    Qwen3 and DeepMiner

    In May 2025, they coauthor Qwen3, which unites thinking and non-thinking modes. In October 2025, they coauthor DeepMiner, a framework for long-horizon search agents that uses high-difficulty training tasks and a dynamic context window.

  6. August 2026

    Qwen-CUA computer use

    In August 2026, they contribute to Qwen-CUA, a computer-use agent that observes screenshots and acts through keyboard and mouse.

  7. September 2026

    Terminal-Universe project lead

    In September 2026, they lead Terminal-Universe, a framework that turns agent trajectories into reusable environments for new tasks and interactions.

  8. As of October 2026

    Researcher in Alibaba's Qwen Team

    They are a researcher in Alibaba's Qwen Team, with interests in language-model intelligence, reasoning, and agent capabilities.

  1. 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.
  2. 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.
  3. 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).
  4. More people shaping AI studied at Massachusetts Institute of Technology than at any other institutionWe estimate where the people shaping AI studied and where their careers led. Massachusetts Institute of Technology leads with 467 former students, including 291 recorded doctorates.

Sources and credits

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

  1. Exploiting Unlabeled Data via Partial Label Assignment for Multi-Class Semi-Supervised Learning ojs.aaai.org. Checked 22 October 2026.
  2. Contrastive Demonstration Tuning for Pre-trained Language Models - ACL Anthology aclanthology.org. Checked 22 October 2026.
  3. Qwen Technical Report arxiv.org. Checked 22 October 2026.
  4. Qwen2 Technical Report arxiv.org. Checked 22 October 2026.
  5. Qwen2.5-Math Technical Report: Toward Mathematical Expert Model via Self-Improvement arxiv.org. Checked 22 October 2026.
  6. Qwen2.5 Technical Report arxiv.org. Checked 22 October 2026.
  7. ProcessBench: Identifying Process Errors in Mathematical Reasoning arxiv.org. Checked 22 October 2026.
  8. The Lessons of Developing Process Reward Models in Mathematical Reasoning arxiv.org. Checked 22 October 2026.
  9. Qwen3 Technical Report arxiv.org. Checked 22 October 2026.
  10. Beyond Turn Limits: Training Deep Search Agents with Dynamic Context Window arxiv.org. Checked 22 October 2026.
  11. Qwen-CUA: Native Computer Use for (almost) Everything arxiv.org. Checked 22 October 2026.
  12. Terminal-Universe: Turning Agent Trajectories into Scalable Terminal Environments arxiv.org. Checked 22 October 2026.
  13. Terminal-Universe | Lab Index labindex.ai. Checked 22 October 2026.
  14. Zhenru Zhang / Zhen-Ru Zhang - Zhenru Zhang zhenruzhang.github.io. Checked 22 October 2026.