will brown
3 Intel
Will Brown is a machine-learning researcher who leads applied research at Prime Intellect, working on open-source research and infrastructure for agentic reinforcement learning, including the Verifiers library and the Environments Hub. He holds a PhD in algorithmic game theory from Columbia University, where he was advised by Christos Papadimitriou and Tim Roughgarden, and studied computer science at the University of Pennsylvania. Before joining Prime Intellect in 2025 he was part of Morgan Stanley's machine-learning research group.
Is this you? Sign in with X to claim this profile.
Intel
Will Brown of Prime Intellect discusses multi-turn RL for multi-hour agents and reacts to Claude 4's focus on longer-running coding, with hosts from Latent Space.
Prime Intellect's Will Brown sketches the roadmap and open questions for agentic RL: what it takes to train an open model that browses, codes and handles long-horizon tasks like OpenAI's o3.
Vals AI's The Bench interviews Prime Intellect's Will Brown on owning versus renting model capability, why skills cannot capture tacit institutional knowledge, token costs approaching salary costs, compute scarcity, benchmark saturation, and sandboxes for agent RL. Transcript covers the first 30 min