Bio

Pengfei Zhang is a PhD student in Computer Science at the University of California, Irvine. His research sits at the intersection of human understanding, multimodal large language models, and multimodal agentic reinforcement learning, with a particular focus on domains where supervision is scarce and the signal is inherently multimodal.

His work spans three directions. He builds VLM-based multimodal agents for human-centric image and video understanding, generation and editing, covering motion representation, action tokenization and identity-preserving multi-turn editing. He works on speech language models and the joint understanding of text, image, video and audio, including post-training methods that align models to pathological and low-resource speech. And he designs personalized LLM agents for healthcare, where knowledge-graph retrieval and long-horizon dialogue optimization are used to tailor conversations to individual patients.

Before UC Irvine, Pengfei received his B.Eng in Computer Science from the University of Science and Technology of China. He has interned at ByteDance, Amazon Web Services and Flawless AI.

He is applying for 2027 New Grad roles on F1 OPT, with expected graduation in Dec 2026, Mar 2027 or Jun 2027.

---

Where to find more

Academic — news, research projects, the full publication list, working experience and awards.

Pengfei’s CV — the full curriculum vitae as a PDF.

Blog — occasional writing.

Reach him at pengfz5@uci.edu.