My work focuses on reward modeling for code LLMs and coding agents, including training scalar reward models based on the Bradley–Terry model for code training data selection, and developing generalist reward models for coding-agent verification.

Under the supervision of Prof. Shing-Chi Cheung, my PhD research focused on Android hybrid programming, including empirical studies of cross-language bugs induced by WebViews and the development of bug detection techniques based on static and dynamic program analysis.

Under the supervision of Prof. Xiaobing Sun, my undergraduate research focused on software evolution and comprehension, leveraging unsupervised NLP techniques based on topic modeling to uncover patterns in software development histories.