Sham M. Kakade

Harvard University

 

Research

I am a full-stack researcher in machine learning and AI, focusing on the engineering, scientific, and mathematical aspects of deep learning. My focus is on developing efficient and practical algorithms for foundation models and real-world AI applications. My current research interests include: (i) optimization with an emphasis on real engineering challenges, (ii) exploring the foundational science and mathematical underpinnings of deep learning, and (iii) advancing the usefulness of LLMs and generative AI. These efforts include understanding how to scale AI architectures and algorithms, taking into account hardware constraints and data composition; exploring the role of RL in language learning and communication and examining the impact of scale on deep learning optimization.

Prospective Students

I am actively seeking students with a strong background in applied deep learning or a keen interest in acquiring these skills. As co-director of the newly-established Kempner Institute, we offer substantial computational resources for cutting-edge research. If you're passionate about driving innovation in these domains or interested in a blend of applied and fundamental research, I encourage you to apply to Harvard!

Book

Reinforcement Learning: Theory and Algorithms

Alekh Agarwal, Nan Jiang, Wen Sun, and I are writing a monograph on Reinforcement Learning. We will be periodically making updates to the draft. Also, see current course CS 6789: Foundations of Reinforcement Learning.

News

Activities and Services