scikit-learn

But be honest with your team about what tests pass actually proves, which on current evidence is less than most people assume.

METR published an analysis in March 2026, where maintainers from three real open-source projects, Scikit-learn, Sphinx, and Pytest, reviewed two hundred and ninety-six AI-generated pull requests thought that already passed SWE-bench verified, a standard benchmark for coding agents.

Roughly half of those test-passing PRs would not have been merged into the real codebase.

The maintainers' reasons weren't nitpicks: core functionality failures, code that broke other parts of the system, and straightforward code quality problems the test suite had no way to catch.

Mm-hmm.

Um, o- the one that probably helped make my career, uh, in a lot of ways was scikit-learn, um-

Okay

...

... just all, all around the world.

Um, and so yeah, it's, it's always had a special place in my heart.

Actually, I think the first presentation I ever gave on Python, um, was at a local meetup on scikit-learn.

Uh, but actually I have, I have a friend now, uh, uh, this was, uh, in the Raleigh area, a friend who, um, that kind of helped s- set the trajectory of his career was coming to that talk and then, uh, getting into data science.

So I think when I look forward to the remainder of this year, I think I want to do two things with GPJax.

I would quite like to close that bridge between GPJax and something like sklearn or GPytorch.

essentially providing a

higher level abstraction of GPJax, which would allow someone to fit a Gaussian process in a few lines of code versus the sort of 20 lines of code we make you write today.

Each year I spent also a little bit of time in Paris, and while being there I met with a couple of groups and organisations for interviews, and one of which is an organisation called Probabel.

But you probably might know them better under the banner of Scikit-Learn, which is the most widely used open-source library for machine learning and AI.

I already talked to the guys at Scikit-Learn when they were still funded by a French research centre.

Since then, they set up a new organisation, well, the one called Probable, and you'll find out how and why it all happened.

The other organization I talked to in Paris is called XWiki, which is also an open source project, and recently created a new branch of a set of web applications called CryptPad.