Bartłomiej Cupiał
Hello there! I'm Bartek, a PhD student at the University of Warsaw, supervised by Piotr Miłoś. My research focuses on reinforcement learning and language agents for long-horizon tasks, with an emphasis on efficient planning, learning from feedback, and multi-agent coordination.
During my PhD, I was a visiting scholar at the UCL DARK Lab under Tim Rocktäschel, where I worked on teaching long-horizon LLM agents when to plan and reason. I've also interned at Sakana AI with Jeffrey Seely, working on decentralized, collaborative multi-agent systems, and at Google under Jarek Mickiewicz, working on teaching LLMs to learn from feedback.
Read about NetHack full moon bug, a story of, by far, the weirdest bug I've encountered in my CS career.