Linqi Zhe

I'm Linqi, a researcher at UPenn working on making reinforcement learning faster and more autonomous. My current focus is an LLM-driven reward-evolution method that pairs reward design with pretrained skill priors, cutting compute by 10x on humanoid locomotion and dexterous manipulation tasks. Alongside that, I'm building at the edge of AI agents and crypto, exploring how autonomous agents can operate on-chain. Ask me about LLM-driven reward design, reinforcement learning for robotics, and AI agents in crypto.