Introduction to AI Agents for Research Workshop

AI agents can now carry out multi-step tasks on their own: writing code, running jobs and reporting back. This introductory, hands-on workshop explains what that means for your research and how to use an agent without losing control of the work. Working on the Kempner cluster, you will brief an agent to perform a simple workflow loop with a small LLM model, set its limits, and supervise it as it works. Guest presenter Ada Fang, Kempner Graduate Fellow from the Zitnik lab, will lead a demonstration of a complex agentic AI workflow. Attendees with no background in computer science, machine learning, or cluster use are welcome to join.

Who can attend?
– Harvard affiliates with Kempner cluster access. Postdocs, graduate students, undergraduates, researchers and staff from any field are welcome. No machine-learning or cluster experience is required.

What will attendees learn from this workshop?
– What an AI agent is, and what it can do for you
– How to ask an agent to do something, by writing a clear brief
– How to restrict what it can access and change, and how to supervise its work
– How to run an agent in a loop, where it proposes, runs, measures, and then keeps or discards changes
– Use cases and applications to different research areas, with discussion and demos

Prerequisites
– An FAS Research Computing account with Kempner cluster access
– Comfort opening a terminal and typing simple commands
– A short pre-work check (about 20 minutes), sent when you preregister
– No prior machine-learning, LLM or cluster experience needed