I am a Ph.D. candidate at Rensselaer Polytechnic Institute (RPI), advised by Mohammed J. Zaki and working closely with Dmitry Krotov at Dynamical Mind (previously worked at IBM Research).
My work is focused on energy-based modeling and Dense Associative Memory, or modern Hopfield networks. I use them for two things: designing neural architectures, and understanding when a generative model memorizes its training data instead of generalizing from it. That second question runs from continuous diffusion models through to language diffusion models, and the same tools carry over to problems outside generative modeling, such as pose estimation in cryo-electron microscopy.
Work from our group on associative memory has been covered by IBM Research (1, 2) and Quanta Magazine. I also keep a blog to contain some thoughts and details about my works.
Recently, I was a summer 2026 research intern at AMD.
