The ARC Evals Team

Daniel Ziegler

Member of Technical Staff

Daniel leads the agent research effort at ARC Evals. Before joining, he spent three years at OpenAI building RLHF and GPT-3, followed by two years at Redwood Research working on adversarial training and other alignment projects. He spent his undergraduate and master’s years at MIT researching formal verification (which he still has a soft spot for), and spends his free time hiking, skiing, and running a large group house.