I am a master student in LTI at Carnegie Mellon University, advise by Prof. Mona Diab, and work with Profs. Maarten Sap, Daniel Fried. My research explores the AI safety, Reasoning, Social Responsible AI, Interpretability, and Humantic AI.

Before coming to CMU, I spent four years at Northeastern University studying Software Engineering. During my undergraduate studies, I completed projects in distributed systems and microservice development.

I’m always enthusiastic about collaborating with researchers from diverse fields. If you’re interested in working together, please don’t hesitate to reach out to me.

Research Interest

AI Safety and Mechanistic Interpretability: Evaluating and mitigating unsafe behaviors in AI systems through both behavioral and latent-space analysis. Investigating the internal representations that govern model behavior to make AI systems safer and more trustworthy, with the long-term goal of designing interpretability-grounded safety benchmarks. I am also currently preparing a related position paper.

Reasoning and Continual Learning: Studying how social and causal reasoning abilities emerge through post-training and reinforcement learning, and uncovering the meta-abilities behind reasoning by probing the internal representations and reasoning trajectories that reinforcement learning amplifies or reshapes.

Human-Centered AI and Benchmarking: Modeling reasoning grounded in human cognition, understanding how computational models of human cognition differ from current AI systems, identifying and mitigating gaps in existing benchmarks that fail to capture realistic human-grounded reasoning, and exploring how insights from human cognitive processes can guide AI systems toward deeper reasoning and better evaluation.

🔥 News

• 2025.07:  Join Fujitsu as a Research Scientist!