Applied researcher at Prime Intellect.
Exploring how language models reason, act, and learn.
I’m an applied researcher at Prime Intellect, working on reinforcement learning and language model agents.
I build environments, design evaluations, and run experiments to understand how models reason, use tools, and learn from feedback.
Reinforcement learning & environments
Build training environments for language model agents, with verifiers that check reasoning and tool use. Run post-training experiments, inspect rollouts, and refine reward signals to help models complete longer tasks and recover when a step goes wrong.
Open models & evaluation
Worked on evaluation tools for open language models, from dataset preparation to reproducible benchmark runs. Compared fine-tuning approaches, investigated model failures, and shared examples that helped developers adapt models to their own tasks.
San Francisco, CARemoteBerlin, DEA few other places I keep things on the internet.