Jiayuan Zhou
jiayuan.zjy[at]gmail.com
Toronto, Canada
I am a Principal Research Scientist at Huawei Canada, leading a research team focused on AI-native software development and intelligent vulnerability management. My team develops cutting-edge AI agent systems for automated software engineering, including the Lingxi Agent framework, which achieved #2 globally on SWE-bench Verified with a 74.6% resolution rate and #1 among open-source models.
I work closely with engineering teams to productionize AI agents for long-horizon software engineering tasks, including cross-file bug fixing, repository-level issue resolution, and end-to-end feature development, with 70-90% measured efficiency gains across multiple product lines.
I received my Ph.D. in Computer Science from the SAIL (Software Analysis and Intelligence Lab), Queen’s University, Canada, under the supervision of Prof. Ahmed E. Hassan and Prof. Shaowei Wang. My dissertation explored extrinsic incentives in open source communities through mining GitHub, Stack Overflow, and Bountysource data.
Research Interests: AI agents for software engineering, LLM-based code generation, multi-agent collaboration, intelligent vulnerability management, mining software repositories.
Publications: My work has appeared in premier software engineering venues including ICSE, ASE, FSE, IEEE TSE, and EMSE. (google scholar).
Industry Experience: Prior to Huawei, I was a Senior Software Engineer at Alibaba Group, where I served as the founding engineer of 1688’s One-Click Dropshipping (一件代发/一键代销) system—a full-stack, end-to-end platform connecting B2B suppliers with millions of Taobao/Tmall merchants for automated order fulfillment and inventory-free retail operations.
I hold 10+ patents in software engineering and AI applications.
news
| Oct 16, 2025 | Our paper was accepted by ICSE’26: “Diffploit: Facilitating Cross-Version Exploit Migration for Open Source Library Vulnerabilities”. See you in Brazil🇧🇷~ |
|---|---|
| Oct 14, 2025 | 🎉 Our developer-experience guided agent solution (Lingxi) achieved 71.2% with Kimi K2 on SWE-bench-verified, ranking #1 🚀worldwide among the open-weight models!! For more details, please check here👀. |
| Aug 25, 2025 | 🎉Our developer-experience guided agent solution (Lingxi) achieved 74.6% with Claude 4 on SWE-bench-verified, ranking #2!🚀 |
| Jan 29, 2025 | Our paper was accepted by TOSEM: “An empirical study on vulnerability disclosure management of open source software systems” |
| Jan 13, 2025 | Our paper was accepted by FSE 2025: “One-for-All Does Not Work! Enhancing Vulnerability Detection by Mixture-of-Experts (MoE)” |