The Causal Mindset Academy/Claude Code for Research

  • $99

Claude Code for Research

  • Course
  • 25 Lessons

Some researchers are not sure where to start with GenAI, others automate full research pipelines. This course aims to help you understand what is possible with GenAI today in research, illustrated with applied examples. Five parts, from first install to fully autonomous research automation.

Trailer

From Zero to Frontier

This course will take you from the foundation to frontier capabilities of GenAI for research, illustrated with many example from my practice.

Contents

This workshop teaches you to master Claude Code, Anthropic's AI coding agent that runs in your computer, and apply it to real research tasks. You'll learn the full system: setup, project architecture, agents, skills, and commands. Then you'll see these concepts applied to several complete, varied, and practical applications for academics and researchers.

1. Setup: Plan Before You Build

Learn how Claude Code reads and reasons about your project so you can configure it properly before executing anything (context windows, plan mode, CLAUDE.md), guided by the Before/During/After framework you'll use all course to use GenAI reliably.

Example: Use Claude Code to understand the structure and content of a research project, clean and reorganize research folders.

Course Overview: Why should you care?
The Before-During-After Framework
Your first session with Claude Code
CLAUDE.md and managing context
Application: Set up a research project

2. One-Shot Execution: From a Single Brief to a Real Deliverable

Prepare a strong brief and let Claude Code produce a complete output in one pass, then verify it. The craft is in the brief, not in babysitting steps.

Example: A full paper drafted from one prompt, and an interactive research website deployed on GitHub Pages.

One-shot execution: Introduction
A simple one-shot: writing a full empirical paper from one brief
An advanced one-shot: when you already have a precise plan
Last example: Paper to a website

3. Tools: Skills, Hooks, Commands, and MCPs

Turn your research expertise into reusable building blocks that Claude Code loads automatically when relevant (skills), call pre-saved prompt (slash commands), define autonomous deterministic rules (hooks), and connect with apps (MCPs).

Example Working skills (academic referee report, applying a naming convention, structuring Latex slides).

Skills: package your instructions once, reuse them
Example: Building the super-referee skill
Optimizing Skills: Personalize, Measure, Improve
Slash commands: pre-saved actions/prompts
Hooks: rules that run automatically
MCP: The Tool That Reaches Outside Your Folder

4. Agent: Sub-agents, Agents Teams, Workflows

Split complex projects across specialized agents that work independently and verify each other's output, solving the problem of a single agent running out of memory (context isolation, scoped permissions, parallel execution).

Example: A course website generated from lecture slides and transcripts using coordinated agents.

Agents: Overview
Sub-agents: hand a task to a worker in its own context
Example: turning a lecture into a web page
The agent view and agent teams
Dynamic workflows: autonomous large-scale tasks

5. Reaching the Frontier: Autonomous Pipeline and Self Improving systems

Design pipelines that run without supervision and get better overnight by testing their own output against your standards (/goal and /loop, self-improving skills, the propose-measure-keep loop).

Example: A full paper replication and extension pipeline that refines itself.

Reaching the frontier: Overview
/goal and /loop slash commands
Self-improving skills
Capstone example: RECAST

6. What now?

This section concludes and opens the discussion to the future of research and how to think about your careers.

What now? Why expertise gets more valuable at the frontier

What Participants Say From my Courses and Talks

⭐⭐⭐⭐⭐
"Sagacity and passion define Quentin's contributions. He makes us better, no matter the circumstances!"

Frédéric Cuguen

Group General Counsel, CW Partners

⭐⭐⭐⭐⭐
"Quentin is the best expert in Causal AI I know. He elegantly combines a deep knowledge of all the subjects relating to statistics, AI and causality, with the most clear explanations for novices and experts alike. I still have his TED talk in mind in which he wrapped up very deep concepts of causality in one, enjoyable and actionable talk.""

Pr. Charles Ayoubi

Expert in technology (AI) and innovation, ESSEC Paris/Harvard Business School

⭐⭐⭐⭐⭐

"Dr Gallea's training provides insights that are counterintuitive and very hard to find in other resources. I strongly recommend to anyone who wants to measure real impact and truly support decision making with data. I can highlight his clarity in explaining the distinction between statistical significance and causal importance, as well as the role of Double Machine Learning in estimating the effects of the one variables on each other."

Marcos Brum

Senior Data Scientist, AI Collaborator, Inc.

⭐⭐⭐⭐⭐

"Combining engaging content, humour, powerful examples and science backed insights, this workshop was fascinating and rewarding. Quentin's master class - "Think causally, Act wisely" - provided our startup executives with a 'statistics first aid kit' and sharpened their tools to deal with bias."

James Miners

Head of startup & innovation programs, Fongit

⭐⭐⭐⭐⭐

"Quentin Gallea is an exceptionally knowledgeable and approachable expert in causal inference. His advice is clear and accessible while remaining rigorous and detailed, and he takes the time to genuinely engage with your questions and research context. His passion for doing causality properly is obvious throughout his work and interactions."

Carlo Maino

Scientific Collaborator, HES-SO

⭐⭐⭐⭐⭐

"I recently took one of Dr. Quentin Gallea's trainings, 'From Prediction to Causation,' and was genuinely impressed. His deep expertise in causality clearly shows, and he explains the real power of causal thinking and how it differs from predictive models in a very clear way. He also highlights common pitfalls, challenges, and risks, helping to deconstruct many assumptions and misconceptions around the fundamentals. Clear, rigorous, and very practical, definitely worth it!"

Camilo Caceres

Data Science/ Machine Learning Expert - Staff Engineer, Mercado Libre

What makes this course different

Built to last

The course is built around concepts and approach, and it illustrates every part with a concrete research example.

More importantly, it gives you a framework for using agentic AI critically. That framework will still be valid years from now, and it is robust to whatever the next model update brings.