LLMs as Autonomous Agents

Harvard Data Science Initative

Explore deploying LLMs as AI agents, focusing on reasoning, planning, and memory with hands-on exercises, techniques, trends, and ethics.

Harvard John A. Paulson School Of Engineering And Applied Sciences

Session Lead: <a href="https://mauriciogtec.com" title="https://mauriciogtec.com" target="_blank" data-msys-clicktrack="0" rel="nofollow noopener noreferrer">Mauricio Tec</a>Session Description: Deploying LLMs as autonomous agents goes beyond text generation––it requires reasoning, planning, and memory. This session explores recent techniques for designing AI agents, including chain-of-thought prompting, memory augmentation, ReAct prompting, and reinforcement learning fine-tuning. Through hands-on coding exercises, participants will gain practical insights into agent design for real-world applications. We will also discuss recent research trends, limitations, and ethical considerations.