Xinran Duan / 段欣然

Multimodal foundation models, agentic AI, and video understanding.