My research interest lies in latent dynamics of world models and reasoning. I investigate the latent dynamics of world models and develop methods to control/explain their behavior and how to endow them with improved reasoning capabilities. A central theme of my research is to understand whether intelligent behavior requires explicit symbolic structure or planning and reasoning can be done purely through optimization in learned latent space. My work leverages ideas from physics informed networks and differentiable logic frameworks to uncover interpretable latent dynamics and improve long horizon prediction. My long term goal is to create world models that are capable of reasoning and planning in latent space while simultaneously developing a framework to understand them.