I’m a Founding AI Scientist at Subconscious. My current research focuses on training and inference for effective, efficient, and long-horizon LLM agents.

During my PhD, I worked with Dr. James Glass’ Spoken Language Systems group at CSAIL. My research focused on augmented and agentic LLMs, with an emphasis on improving LLM agents’ retrieval and tool-use capabilities. I also worked on knowledge-grounded AI, aiming to ground models in factual and structured knowledge.

I hold a Master’s degree from MIT, where I worked on automatic fact verification and speech synthesis under Dr. Glass. I received my Bachelor’s degree from National Taiwan University, working on speech and language understanding with Prof. Hung-Yi Lee and Prof. Lin-Shan Lee, and computer vision with Prof. Yu-Chiang Frank Wang.