Guoqing Wang

Guoqing Wang

Fourth-year Ph.D. student at CS@PKU, interested in AI4SE(code generation, test generation, and software maintainance), Large Language Models (LLMs).

Guoqing Wang (王国庆)

Guoqing Wang is a fourth-year Ph.D. student at CS@Peking University, co-advised by Prof. Dan Hao and Prof. Zeyu Sun. He received his bachelor’s degree from Harbin Institute of Technology (HIT) in 2022.

His research resides at the intersection of AI for Software Engineering (AI4SE) and Trustworthy/Truthful AI. He is dedicated to building reliable and controllable Large Language Model (LLM) systems, with a focus on enhancing their performance in software development and ensuring their factual integrity in knowledge-intensive tasks.

🔥 I am currently on the academic/industrial job market for research positions. Please feel free to reach out for potential collaborations!

📢 News

  • [April 2026] Our paper PurifAI: Detecting and Fixing Search-Induced Distortions in Web-Augmented LLMs was accepted by SIGIR 2026 (Full Paper)! I will be attending SIGIR 2026 in Melbourne, Australia. Let’s connect!
  • [April 2026] Our paper From Greedy Steps to Global Optimization: Learning Sequential Test Suite Generation was accepted by ISSTA 2026 (Full Paper)!

🔬 Research Pillars

My research agenda is driven by the goal of making LLMs both effective for specialized domains and trustworthy in open environments.

1. AI for Software Engineering (AI4SE)

Focuses on leveraging foundation models to automate and optimize the software lifecycle:

  • Automated Testing & Debugging: Designing techniques for unit test generation and fault localization (e.g., bug-inducing commit localization).
  • Code Intelligence: Improving the reliability and controllability of code generation and editing through prompt engineering and alignment training.

2. Truthful & Trustworthy AI

Focuses on the reliability and factual integrity of LLMs, especially in RAG (Retrieval-Augmented Generation) scenarios:

  • Hallucination Mitigation: Proactively detecting and fixing “search-induced distortions” where external noise overrides internal verified knowledge.
  • Knowledge Alignment: Ensuring LLM outputs remain consistent with trusted knowledge bases (TKB) in high-stakes domains (Legal, Medical, Enterprise).

You can find my full list of publications on Google Scholar. For more details about my experience, please refer to my [CV] or contact me via email.