Information at every scale.

Learning, reasoning, and the information that makes intelligence possible.

I work on AI and large language models at Zyphra. My interests include how models learn and reason, with a focus on the training methods that shape their capabilities.

My background in information theory informs how I think about learning, representations, and the limits of inference. Areas of interest include pre-training, post-training, and reinforcement learning with verifiable rewards.

My AI/LLM work is a current research direction. Public papers and project links will be added as they become available.

The structure of quantum information and the limits of what we can learn.

What can we learn about a quantum system, and how efficiently can we describe, transform, and communicate its information? My work explores these questions through quantum information theory and mathematical physics.

Topics include quantum state estimation, state merging, cloning, entropy and divergences, and quantum minimum description length. Recent projects also use Lean to make mathematical arguments formally verifiable.

RELATED WORK

Spacetime, black holes, and the information hidden in geometry.

Black holes bring quantum theory and gravity into direct conversation. I study what information-theoretic ideas can tell us about black hole entropy, holography, and the emergence of spacetime.

My research connects the quantum de Finetti theorem, one-shot entropies, and free probability with the black hole information problem. I have also worked on causal sets, quasilocal mass, and the geometry of small causal diamonds.

RELATED WORK