My research lies at the intersection of  statistical social science and artificial intelligence, aiming to advance trustworthy AI that is both technically robust and socially beneficial.  To achieve this, I explore two key directions:  developing frameworks to maximize the reliability of large language models, including addressing reward overoptimization and enhancing model steerability, and developing self-improving methodologies that enable AI systems to autonomously explain and refine their reasoning without extensive human labor.

Please feel free to read through my cv, and contact me. Thank you!