Vision: Intelligent machines shape how we perceive the world: how we model space, how we reason about motion, and how we make decisions that affect other people. By changing the representations and objectives that underlie these systems, we change the structure of machine reasoning itself. How do we design machines that are both physically grounded and socially aligned? My research advances models and architectures that unify world structure with human values. In doing so, I aim to address foundational questions: How can we build representations that capture geometry, dynamics, and causality rather than surface correlations? How do we embed uncertainty, ethics, and cultural context directly into learning systems? And how should intelligent systems be designed if they are to operate not only accurately, but responsibly within society?

---