COURSE STRUCTURE
Each block runs over two weeks. You learn at your own pace, then come together for a live session with both instructors.
Week A
Pre-recorded videos and hands-on notebooks. Watch, code, and experiment on your own schedule.
Week B
Python exercises, and open discussions on this platform with other members of the cohort and instructors.
End of each block
Every two-week block ends with a live debrief (planned around 5pm CET for now, recording available).
WHY?
Academic rigor meets industry-grade practice. Quentin from measuring GenAI and AI effects, Matheus from applied data science in a large fintech.
We illustrate class concepts with our applied business experience. Every method is paired with real use case.
A special space to discuss for the group across all 16 weeks on this platform plus 8 live debriefs with both instructors keeps you connected and maximize the clarity and applicability of the concepts.
Causal vs. predictive inference, DAGs, potential outcomes, good/bad controls, a word on causal discovery.
Design, analysis, and heterogeneous treatment effects.
Identification with observed control variables and Double/Debiased Machine Learning.
Leveraging within-unit variation for credible causal estimates.
Canonical Diff-in-Diff, Two Way Fixed Effects Diff-in-Diff, staggered adoption, and event study designs.
Sharp and fuzzy RD designs.
IV intuition, identification assumptions, and modern practice.
Sensitivity analysis (Cinelli & Hazlett) and CATE estimation for decisions.
WHO?
Built for data scientists, ML engineers, and applied researchers who already work with data and want to learn and apply causal inference concepts in business.
Working knowledge of basic statistics (hypothesis testing, p-values, random variables) and in particular of linear regression. Comfort with Python for data science. Familiarity with machine learning is helpful but not required.
YOUR FIRST INSTRUCTOR
Senior Data Scientist at Nubank and author
Matheus Facure was an economist and Senior Data Scientist at Nubank, the biggest FinTech company outside Asia, where he applies causal inference to real business decisions at scale, across credit, pricing, marketing, and cross-sell.
Applied causal methods to million-customer-scale business problems at Nubank, including automated credit and interest decisioning, marketing budget optimization, and cross-sell.
Author of Causal Inference for the Brave and True (open-source, widely used in the practitioner community).
Author of Causal Inference in Python: Applying Causal Inference in the Tech Industry.
YOUR SECOND INSTRUCTOR
With deep expertise and a passion for causality, Quentin brings scientific rigor to answering causal questions and applying Causal AI to real-world problems, particularly to help deploy reliable and safe AI strategies.
Delivered workshops for billion-dollar companies (including Google)Advised C-suites and data leaders worldwide on causal inference and AI impactTrained 15,000+ students and professionals across industriesPublished research in top scientific journals
Speaker at leading international events, including:
TEDx
Causal Data Science Meeting
Applied Machine Learning Days
National Association for Business Economics
Organizer of The Causal Summit
Author of The Causal Mindset Handbook
FAQ section
You've got questions. We've got answers. If you still have questions, you can contact us directly (e.g. quentin[at]quentingallea.com)
Yes, and honestly this is who the course is built for. If you already work with these methods, three things will move the needle for you:
1. You will learn how to do it right in practice, with the latest methods. We start from the foundations, but the point is not to re-teach A/B testing. It is to show you the cutting-edge version of each method and how to apply it correctly on real problems. Everything we teach has been pressure-tested in our own practice, not just read off a textbook.
2. You get the practitioner insights that are rarely taught. Most courses stop at the off-the-shelf algorithm. We go past it. For example, Matheus shares how he computes heterogeneous treatment effects efficiently under real business constraints, not just by calling a library. And Quentin covers the Type III error: how a perfectly measured causal effect can still lead to a bad decision in practice, and what to do about it. These are the lessons that usually only come from years on the job.
3. You join a room of peers and get direct access to us. You will connect with other experienced practitioners, trade war stories, and compare approaches. And you can bring your own applications and ask us specific questions about them. That feedback loop is hard to get anywhere else.
If you are already comfortable with the basics, you will spend your time on the parts that actually make you better: doing it right, avoiding the traps, and applying it to your own work.
No, as long as you have the pre-requisites (working knowledge of basic statistics (linear regression) and hypothesis testing, as well as comfort with Python for data science.
The course starts from the foundations, then builds up one block at a time. Each method rests on the ones before it, so by the time we reach the harder material like double machine learning or staggered difference-in-differences, you have already seen every idea it depends on.
By the end you can run a causal analysis end to end on your own data, from framing the question to defending the estimate.
You will be able to design and analyze experiments, including heterogeneous treatment effects, and reach for the right observational method when a clean experiment is not an option.
We will present you cutting edge methods, considering the usual constraints in business (sample size, time frame etc.) and discuss all the subtle concepts that might lead to costly mistakes if misunderstood.
The goal to allow you to do all of this yourself, on your own problems in a reliable way with the best methods available.
Plan for about 4 hours a week over 16 weeks. The program is built as 8 two-week blocks, and each block splits the work in two. Week A is self-paced: pre-recorded videos and hands-on Python notebooks you work through on your own schedule, around your job. Week B is where you apply it, with exercises, group discussion, and one 60-minute live debrief.
The students are working professionals from all over the world. Hence, we build a course that is flexible enough to fit your busy schedule (Matheus and I, have both families, and know what it means to balance all this (or at least try to)).
If you miss the live session, you can watch it later.
The two-week rhythm is there on purpose. It gives you slack to catch up when a week gets away from you.
Yes, and this is already the case for some participants covering the cost as professional development. If that is your situation, we can issue an invoice made out to your company so your finance or L&D team can process it directly.
If it helps to make the case internally, reach out before you enrol and we can give you what you need, whether that is an invoice, the course outline, or a short summary of what you will be able to do by the end. Just get in touch and we will sort out the details.
Yes, at the end of the program, if you went through all the lessons, you'll receive a certificate of completion.
This is just conditional on going through all the sessions. No other requirement (like a graded assignment etc.).
You have a lifetime access to the course. Hence, you will be able to review all the documents, case studies and videos later.
Once you enroll and the course starts, you will have access to a private space with other participants. This will allow you to network, share experiences and challenges, learn from others.