Haojun Yan

World Models · Robot Learning · 3D/4D Vision

I am a World Model Research Intern on the Genie Team at AGIBOT, working on evaluating robot policies with generated video rollouts. I also work with Prof. Ayush Tewari at the Visual Structure Learning Lab, University of Cambridge, on dynamic scene reconstruction.

My research interests lie at the intersection of world models, robot learning, and 3D/4D vision. My work spans reconstructing dynamic environments from sparse observations, learning bimanual manipulation from human demonstrations, and evaluating robot behavior in learned simulators.

I studied Machine Learning and Machine Intelligence at the University of Cambridge (MPhil, 2025–2026), and received my B.Eng. in Computer Science and Technology from Beihang University in 2025. Previously, I interned at LatenCore AI and Galaxea AI.

I am seeking PhD opportunities in world models, robot learning, and 3D/4D vision. Please feel free to get in touch.