Hi there, I’m Lei Zhang. 👋

I’m a Ph.D. student at the University of Chinese Academy of Sciences, Beijing, China, with research interests in Agentic Coding, Agentic Infra, and Agentic Reinforcement Learning.

I recently completed an internship at Alibaba Qwen Group (Aug. 2024 – Mar. 2026), where I was responsible for code evaluation pipelines for the Qwen2.5 and Qwen2.5-Coder series, and for the agentic coding infrastructure and training data synthesis for the Qwen3 and Qwen3-Coder series.

I believe that almost everything in the modern world is built on software — and software is built on code. A model that truly understands and writes code is not just a tool for developers; it is foundational to nearly every future application of AI. This conviction drives my focus on advancing model capabilities in agentic coding scenarios, where I see the greatest potential for real-world impact.

Academic Achievements

  • Ph.D. in Computer Application Technology, University of Chinese Academy of Sciences, 2021-present
  • B.Sc. in Computer Science, University of Chinese Academy of Sciences, 2017-2021

Projects

  • Qwen3 and Qwen3-Coder, Alibaba Qwen Team

    Core Contributor

    Responsible for the design and implementation of the infrastructure for model training and evaluation workflows for long-horizon agent tasks; responsible for the coding agent data synthesis for model training.

  • Qwen2.5 and Qwen2.5-Coder, Alibaba Qwen Team

    Contributor / Core Contributor

    Responsible for integrating and conducting code evaluation (EvalPlus, Multiple, LiveCodeBench, BigCodeBench, CRUXEval, Aider, etc.) for iterative model development.

Technical Reports

  • Qwen3-Coder-Next Technical Report
  • Qwen3 Technical Report
  • Qwen2.5-Coder Technical Report
  • Qwen2.5 Technical Report

Selected Publications

See full list →

  • MegaFlow: Large-Scale Distributed Orchestration System for the Agentic Era, [Preprint]

    Lei Zhang, Mouxiang Chen, Ruishen Cao, Jiawei Chen, Fan Zhou, Yiheng Xu, Jiaxi Yang, Liang Chen, Changwei Luo, Kai Zhang, Fan Yan, Jiachun Shen, Jiajun Zhang, Zeyu Cui, Feng Hu, Junyang Lin, Binyuan Hui, Min Yang

  • SWE-Universe: Scale Real-World Verifiable Environments to Millions, [Preprint]

    Mouxiang Chen, Lei Zhang, Yunlong Feng, Xuwu Wang, Wenting Zhao, Ruisheng Cao, Jiaxi Yang, Jiawei Chen, Mingze Li, Zeyao Ma, Hao Ge, Zongmeng Zhang, Zeyu Cui, Dayiheng Liu, Jingren Zhou, Jianling Sun, Junyang Lin, Binyuan Hui

  • SWE-RM: Execution-free Feedback for Software Engineering Agents, [ICLR'26, CCF-A]

    Jiachun Shen, Binyuan Hui, Jiawei Chen, Lei Zhang, Jiaxi Yang, Junyang Lin, Junxian He

  • SWE-Flow: Synthesizing Software Engineering Data in a Test-Driven Manner, [ICML'25, CCF-A]

    Lei Zhang, Jiaxi Yang, Min Yang, Jian Yang, Mouxiang Chen, Jiajun Zhang, Zeyu Cui, Binyuan Hui, Junyang Lin

  • Hierarchical Context Pruning: Optimizing Real-World Code Completion with Repository-Level Pretrained Code LLMs, [AAAI'25, CCF-A]

    Lei Zhang, Yunshui Li, Jiaming Li, Xiaobo Xia, Jiaxi Yang, Run Luo, Minzheng Wang, Longze Chen, Junhao Liu, Min Yang

  • Marathon: A Race Through the Realm of Long Context with Large Language Models, [ACL'24, CCF-A]

    Lei Zhang, Yunshui Li, Ziqiang Liu, Jiaxi Yang, Junhao Liu, Longze Chen, Run Luo, Min Yang

Research Interests

  • Agentic Coding
  • Agentic Infra
  • Agentic Reinforcement Learning

Contact

Feel free to reach out via jameszhang2880@gmail.com for collaboration or questions.