I am a researcher based in Toronto. My mission is to develop mathematical foundations for understanding how neural networks learn structured representations of the world, with emphasis on discovering and interpreting the geometric symmetries that emerge during active inference. My research combines algebraic geometry, differential geometry, bayesian inference, and mechanistic interpretability techniques to decode the internal organization of deep learning systems—uncovering how networks develop compositional world models and how these structured representations are shaped during perception and planning. This work aims to bridge theoretical understanding of representational geometry with practical tools for interpreting and aligning advanced AI systems.