Early Training

My training started with a PhD in Neurobiology at Duke University. I studied how the cerebellum, the part of the brain involved in motor control, plans and controls eye movements under the mentorship of Stephen Lisberger and Marc Sommer.

This work sparked my interest in control theory and systems identification—the use of mathematical models to describe how feedback systems operate. These ideas still shape my thinking.

Postdoctoral Research

As a postdoc at NYU’s Center for Neural Science in Tony Movshon’s lab, I shifted to computational vision. I combined behavioral and neurophysiological methods with computer vision tools to investigate how the brain encodes and decodes visual information.

Leading translational projects with clinical ECoG arrays introduced me to real-world, imperfect settings.

Current Approach

I continue to approach applied AI and computer vision systems as a psychologist would study perception. I see them as parts of a broader cognitive architecture of memory, reasoning, and decision-making under uncertainty.

Working in safety-critical environments constantly reminds me of the gap between idealized models and reality. This reinforces my need to move between theory and practice.

For that reason, I contribute to open-source implementations of psychological, visual, and—more recently—risk models. Writing software is a way for me to formalize assumptions, interrogate their limits, and carry insights from theoretical work back into deployed systems. It also provides a concrete mechanism for refining reasoning through implementation.