About me

I am a PhD student in Software Engineering at the University of Technology Sydney (UTS), supervised by Prof. Yulei Sui. I expect to graduate in mid-2027 and am looking for research positions in AI safety.

I was a Research Assistant (Mentor: Ting Su) at East China Normal University (2022-2023) and a Visiting Scholar at Ant Group, Hangzhou, China (2022-2023). I got my M.S. in Software Engineering from East China Normal University (ECNU), Shanghai, China (2018-2021).

News

• Oct 2026: New blog post: Transformer Notes, using Gemma 2 27B as the running example (in Chinese).
• Oct 2026: Our paper on synthesizing CodeQL queries from seed patches was accepted to ICSE 2027. The generated queries found 49 previously unknown bugs in the Linux kernel.

My research interests are in:

• AI safety through the lens of program analysis. Many safety mechanisms, such as guardrails, prompt-injection detectors, AI-control monitors and reward verifiers, are judges that make false positives and false negatives. I want to understand why they fail, building on my work on finding and understanding defects in static analyzers.
• Static analysis: LLM-assisted static analysis that turns bug-fix patches into reusable CodeQL queries for the Linux kernel (ICSE’27), and alias specification generation for library APIs (TOSEM’25).
• Fuzzing and software testing: fuzzing compilers and program analyzers to find their false positives and false negatives (FSE’24), diagnosing automated GUI testing tools (FSE’23), and test generation via dynamic symbolic execution (APSEC’21, SCP’21).

Publications

1. Seed Patches to CodeQL Queries: Compositional Synthesis with Proof-of-Concept Validation.
   Weigang He, Xiao Cheng, Shuangxiang Kan, Wei Li, Xingrui Yu, Ivor Tsang, Yulei Sui
   IEEE/ACM International Conference on Software Engineering (ICSE), 2027. To appear.

2. Spectre: Automated Aliasing Specification Generation for Library APIs with Fuzzing.
   Shuangxiang Kan, Yuekang Li, Weigang He, Zhenchang Xing, Liming Zhu, Yulei Sui.
   ACM Transactions on Software Engineering and Methodology (TOSEM), 2025.

3. Finding and Understanding Defects in Static Analyzers by Constructing Automated Oracles.
   Weigang He, Peng Di, Mengli Ming, Chengyu Zhang, Ting Su*, Shijie Li, Yulei Sui
   ACM Joint European Software Engineering Conference and Symposium on the Foundations of Software Engineering (FSE), 2024.

4. Automata-based Trace Analysis for Aiding Diagnosing GUI Testing Tools for Android.
   Enze Ma, Shan Huang, Weigang He, Ting Su*, Jue Wang, Huiyu Liu, Geguang Pu, Zhendong Su
   ACM Joint European Software Engineering Conference and Symposium on the Foundations of Software Engineering (FSE), 2023.

5. Data flow testing for PLC programs via dynamic symbolic execution.
   Weigang He, Xia Mao, Ting Su, Yanhong Huang, and Jianqi Shi.
   28th Asia-Pacific Software Engineering Conference, APSEC 2021, Taipei, Taiwan, December 6-9, 2021, pages 152–160. IEEE, 2021.

6. Automated test generation for iec 61131-3 st programs via dynamic symbolic execution.
   Weigang He, Jianqi Shi, Ting Su, Zeyu Lu, Li Hao, and Yanhong Huang.
   Science of Computer Programming, 206:102608, 2021.

7. Safety verification of iec 61131-3 structured text programs.
   Jiawen Xiong, Xiangxing Bu, Yanhong Huang, Jianqi Shi, and Weigang He.
   IEEE Transactions on Industrial Informatics, 17(4):2632–2640, 2020.

Workshop

1. QLM: CodeQL Query Synthesis with Compositional PoC Validation.
   Weigang He, Xiao Cheng, Shuangxiang Kan, Wei Li, Xingrui Yu, Ivor Tsang, Yulei Sui
   Journal Ahead Workshop (JAWs), co-located with ICSE 2026. Presentation; an earlier version of the ICSE’27 paper.

Services

I serve as a reviewer for IEEE Transactions on Software Engineering (TSE), ACM Transactions on Embedded Computing Systems (TECS) and IEEE Transactions on Reliability (TR).

Teaching Experience

Tutor, 41128 Software Analysis Studio, UTS, 2024-2025. Tutor and lab demonstrator, 41184 Secure Programming and Penetration Testing, UTS, 2025

[Last updated on Oct 2026]