Agentic Research Course¶
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 + 15 min Q&A
- Schedule: Wednesdays at 9:00 AM Pacific / 12:00 PM Eastern / 4:00 PM UTC
- Dates: April 8 -- June 9, 2026
- Zoom: Join via Zoom
- Nextcloud Talk: Join via Nextcloud (alternative for those without Zoom access)
- Office Hours: Fridays 9-10 am PT, Join via Zoom
- Recordings: Uploaded to YouTube after each session
- Community: Discord (OpenScience Collective)
- Plugin: research-skills (free, open source)
- Source: GitHub
- License: CC-BY-4.0
Schedule¶
| Date | Session | Topic |
|---|---|---|
| Wed April 8 | Week 1 | Git, GitHub, and the Command Line |
| Wed April 15 | Week 2 | Setting Up Claude Code for Research |
| Tue April 21 | Week 3 | Project Management with AI |
| Wed April 29 | Week 4 | CI/CD and Code Quality |
| Wed May 6 | Week 5 | Literature Search and Review |
| Wed May 13 | Week 6 | Grant Proposal Writing |
| Wed May 20 | Week 7 | Manuscript Preparation and Peer Review |
| Wed May 27 | Week 8 | Scientific Figures |
| Wed June 3 | Week 9 | Neuroinformatics: Standards, Sharing, and Credit |
| Tue June 9 | Week 10 | Building Your Own Plugins |
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: A Real Research Project¶
Every workflow in this course is demonstrated against an actual open research question, not a toy dataset. The practicum asks: is there a group-level electroencephalography (EEG) signature locked to movie shot changes, and does it depend on what is in the shot?
Concretely, we compare event-related spectral perturbation (ERSP) in the first 500 ms after shot onset between shots in the Pixar short "The Present" that open with the boy visible against shots that open with the puppy visible. The contrast isolates content-driven brain responses (animacy, social attention) from low-level visual-onset effects that every shot change shares.
The dataset is the published Healthy Brain Network (HBN) EEG Release 3 on NEMAR and OpenNeuro: 183 children and adolescents watching The Present, 128-channel high-density EEG, formatted with Brain Imaging Data Structure (BIDS) and annotated with Hierarchical Event Descriptors (HED). Shot-level annotations come from the EventFormer project (Shirazi lab): 56 shots with frame-level descriptions, log-luminance-ratio, and content flags (has_boy, has_puppy).
This is not a reproduction of an existing result. The boy-versus-puppy ERSP contrast has not been published for this dataset, and the animacy-of-opening-shot question is genuinely open. The deliverable is a reproducible pipeline (BIDS import -> cleaning -> adaptive mixture independent component analysis (AMICA) -> IClabel -> shot-aligned epoching -> ERSP -> cluster-level statistics) plus a publication-quality figure of the group contrast. The instructor drives the analysis live each week using EEGLAB plus matlab-mcp-tools (Claude Code drives MATLAB).
Students are welcome to bring their own research projects and apply the same workflow in parallel. The scaffolding transfers: structured problem definition, epic/phase decomposition, plan mode, pull request review.
Practicum repository
Starter materials (the structured problem brief and the shot-events table) live at sessions/week-03/practicum/ in the course repo. The practicum itself is built live during the Week 3 session at OpenScience-Collective/agentic-research-practicum.
Instructor¶
Seyed Yahya Shirazi, Ph.D. Assistant Project Scientist, Swartz Center for Computational Neuroscience, UC San Diego