Hello! I'm Rustem, and this is my personal page.

My research focuses on AI alignment and scalable oversight. I'm currently a Research Scholar at MATS in London, working with Prof. Shi Feng on the realism of model organisms of misalignment, the science of generalization, applying invariance to alignment problems, and consistency training. I'm also interested in developing counterpart domain randomization as a method for AI control and in curriculum learning. Previously, I was a Research Engineer at Palisade Research, where I evaluated LLM agents in offensive security settings and studied how they misbehave.

Outside research, I care a lot about music. One day I'd love to build an AI, or just a tool, that helps compose great new music, with the richness of the Romantic and Baroque traditions.