Short bio: I am an assistant professor and Zhu Chair at UCSB CS.
I completed my Postdoc at UC Berkeley as part of the RDI center and BAIR Lab, under the mentorship of Prof.
Dawn Song.
I received my Ph.D. from Penn State with Prof. Xinyu Xing and my master's
degree from Shanghai Jiao Tong University.
Outside of research, I was a founding member and led the Agent Security R&D at Virtue AI, which was acquired by Meta.
Research Interests
My current research centers around:
LLM agents and agentic RLs, including environment constructions, agent design, and specialized model training
Agentic inference and model serving, e.g., KV cache design, serving new models
LLM reasoning and post-training for coding, including general RL post-training techniques and specific training recipes for improving coding-related capabilities
Representive research project: OpenSage - Next-generation Agentic AI.
My previous works studied the trustworthiness of DRL and DNN via statistical modeling (e.g., Deep GP, Gaussian process
models) and learning (e.g., variational inference, MCMC, empirical bayes),
and designing DNNs for security applications with a focus on noisy learning and OOD.
Honors and Awards
Academic Senate Faculty Research Award, UCSB, 2025
