About Guanqun (David) Yang
I’m Guanqun (David) Yang, an LLM researcher. I study how large language models and AI agents can test software, secure it, and check the work of other AI systems.
What this newsletter covers
Each post explains one piece of research in plain language: the problem, the method, the numbers, and what it means for a team that builds or buys AI. The newsletter has three sections:
Research Explained: one peer-reviewed paper per post.
Field Notes: shorter notes on LLMs, AI agents, and software engineering.
Announcements: new papers, talks, and events.
Research
Security vulnerability management: FewVuln (MLHat at KDD 2021) and PatchHolmes (AACL-IJCNLP 2026).
Testing AI models: TestAug (COLING 2022) and HateModerate (Findings of NAACL 2024, WOAH 2024 Outstanding Paper Award).
AI coding agents and agent skills: WatchPoint (arXiv, 2026) and SkillSeek (AACL-IJCNLP 2026).
Recommender systems: Sequential Modality Dropout and DualSpectralCF (CIKM 2026).
Explainable claim verification with several LLMs: ART (Findings of EACL 2026), from a research internship at Capital One.
Background
Stevens Institute of Technology, 2020 to 2026: research on scalable software testing and analysis with NLP methods, advised by Xueqing (Susan) Liu. Read the dissertation (PDF).
Capital One, summers of 2024 and 2025: research internships on LLM fine-tuning for chatbots and on multi-LLM systems for legal validation.
UCLA, M.S. in Electrical and Computer Engineering, 2019.
Northeastern University (China), B.Eng. in Electrical Engineering, 2017.
Find my work
guanqun-yang.com · Google Scholar · LinkedIn · GitHub · ORCID · guanqun.yang@outlook.com

