AI-Powered Research Workflows¶
A 10-week live course teaching researchers how to use Claude Code and AI coding agents for academic research workflows, hosted by the Open Science Collective.
Format¶
- Sessions: 30-45 min live on Discord Stage + 15 min Q&A
- Schedule: Weekly (Wednesdays)
- Recordings: Uploaded to YouTube after each session
- Community: Discord (OpenScience Collective)
- Plugin: research-skills (free, open source)
- Source: GitHub
- License: CC-BY-4.0
Syllabus¶
| Week | Topic | Materials |
|---|---|---|
| 1 | Git, GitHub, and the Command Line | Foundations + landscape overview |
| 2 | Setting Up Claude Code for Research | First AI session |
| 3 | Project Management with AI | Epics, sprints, worktrees |
| 4 | CI/CD and Code Quality | GitHub Actions, linting |
| 5 | Literature Search and Review | opencite, synthesis |
| 6 | Grant Proposal Writing | NIH/NSF proposals |
| 7 | Manuscript Preparation and Peer Review | IMRAD, formatting |
| 8 | Scientific Figures | matplotlib, react-pdf |
| 9 | Neuroinformatics | BIDS, PsychoPy, LSL |
| 10 | Building Your Own Plugins | Skills, agents, hooks |
Prerequisites¶
- A computer with terminal access (macOS, Linux, or Windows Subsystem for Linux (WSL))
- A GitHub account (free)
- No prior coding experience required for Weeks 1-2
Practicum¶
The course includes a running practicum analyzing Healthy Brain Network (HBN) EEG data for movie shot-change events. The instructor demonstrates each workflow live using EEGLAB + matlab-mcp-tools (Claude Code drives MATLAB). Students are encouraged to bring their own projects.
Instructor¶
Seyed Yahya Shirazi, Ph.D. Assistant Project Scientist, Swartz Center for Computational Neuroscience, UC San Diego