2026 workshop program
Introduction to Agentic Coding for Researchers
The work
Abstract
This tutorial is a practical introduction to coding agents: AI assistants that, unlike a chat window, work directly on your machine. Instead of you copy-pasting code and errors back and forth, the agent reads your files, runs your code, sees the errors, and corrects accordingly. We’ll start with a hands-on setup on your laptop and an overview of how these tools work: a language model equipped with a few simple tools, mainly viewing and editing local text files. The rest of the session will cover how to work with agents effectively: directing them, staying in the loop, and verifying what they produce. We’ll pay particular attention to a failure mode common in research code: results that look reasonable but are wrong. Participants are encouraged to bring a laptop and follow along.
Requirements: GitHub Repo