Daniel Ritter

About

Daniel Ritter is a visiting Ph.D. student from Cornell University working with Kianté Brantley. He works at the intersection of reinforcement learning and generative modeling, and designs algorithms for efficient and effective reinforcement learning on large pre-trained generative models. In the past, he also worked in the Marks Lab at Harvard Medical School, applying generative models to problems in computational biology. Before that, he received his Masters in Computer Science from the University of Oxford, and his Bachelors in Computer Science and Political Science from Brown University.

Research Focus

Daniel Ritter’s research centers on designing novel, theoretically justified reinforcement learning algorithms, with a primary interest in algorithms that can be efficiently applied to pretrained generative models. His current work focuses on improving the policy gradient algorithms used for post-training in large language models and on improving the sample efficiency of LLM post-training by developing higher performing off-policy reinforcement learning algorithms.