Marc Gong Bacvanski

I'm currently a PhD student at MIT, where I work on analog hardware for intelligence. The field of Machine Learning has given us a mathematical formulation for learning, and today GPUs evaluate that math via algebra. I want to build Learning Machines: physical systems that embody learning rather than merely simulating it.

I'm drawn to attention, balance, craft,
light, fractals, contrast,
wonder, mischief, and the surreal—
especially when shared.

Currently

I am interested in building the full stack of intelligence: hardware and algorithms, together.

Recently

January 2026: Invited talk — Aspen Center for Theoretical Physics, Theoretical Physics for Artificial Intelligence. Slides.

December 2025: ArXiv preprint of Dense Associative Memories with Analog Circuits

October 2025: ArXiv preprint of On Biologically Plausible Learning in Continuous Time