Research

Work at the frontier.

Papers, experiments, and open questions in AI and machine learning.

Papers in progress. Check back soon — or watch the GitHub for experiments that might turn into something.

Language Model Architectures

Exploring efficiency in transformer variants — mixture of experts, dynamic depth, and what happens when you take these ideas to very small scale.

AI Agents & Tool Use

Building and studying agents that interact with the real world through tools — browsers, CLIs, APIs. How do you make them reliable?

Local & Private AI

Making capable AI systems that run entirely on-device. Privacy, latency, and the question of what you actually need a cloud for.