Build Your Own RA (Research Agent)

Introduction to Agentic Coding, Git, and Custom Research Workflows

28 June 2026, 8:35 AM – 12:00 PM (local time)

Faculty

  • Jacob Jameson, MS – PhD Student, Health Policy/Decision Sciences, Harvard University, United States

Course Overview

The future of research is changing rapidly. AI-powered coding tools are transforming how researchers write, debug, and maintain code. Researchers are shifting from writing every line to directing, reviewing, and iterating with intelligent agents. For health researchers using simulation models, analyzing large datasets, or developing analytic pipelines, these tools offer dramatic gains in productivity and reproducibility. However, effective work with AI agents requires new skills, including learning how to prompt, customize, and maintain control over what the agent produces.

This hands-on course introduces researchers to agentic coding using AI tools such as Claude Code to accelerate research programming. Participants will learn the foundations of agentic workflows, including how to prompt effectively, customize agents for domain-specific tasks, and build reusable templates.

The course also integrates git and GitHub fundamentals, reframing version control as an essential operating system for safe collaboration with AI agents. Participants will direct agents to work on branches, review their changes through pull requests, and merge only approved changes to maintain control of their codebase.

The course culminates in a hands-on build session where participants begin developing a custom agentic workflow tailored to their own research. Participants will leave with practical skills and a prototype workflow they can continue developing.

Learning Objectives

Participants will learn how to:

  • Describe how agentic coding tools are transforming research workflows and their applications in research
  • Apply git fundamentals—commits, branches, and pull requests—to manage agent-assisted code development
  • Develop a prototype agentic workflow tailored to a personal research project

Course Format

This course combines short lectures, demonstrations, and hands-on exercises.

Participants will be introduced to the foundations of agentic coding and version control, including git and GitHub workflows for collaboration with AI agents. The session includes demonstrations of branching, pull requests, and code review processes, as well as guidance on customizing agents for research tasks.

The course concludes with a hands-on build session where participants begin developing their own agentic workflow with facilitator support.

Participant Requirements

Participants should have basic familiarity with R or Python (no advanced programming experience required).

Participants must bring a laptop with git installed, a GitHub account, and Claude Code installed. A detailed setup guide will be provided in advance. Participants are encouraged to bring a research task or project in mind, such as a simulation model, data pipeline, or analysis script.