François Goupil — VP Services and co-founder @ Probabl - scikit-learn core contributor
VP Services and co-founder @ Probabl - scikit-learn core contributor
François Goupil ranks #565 of 18,341 LinkedIn creators in Computer Software, and is a standout voice in France. They have 8.6K followers and published 22 posts in the last 30 days at a 1.7% average engagement rate.
- 8.6K followers
- 22 posts / 30d
- 1.7% avg engagement
- 222 follower growth / 30d
The roast
François spends his days claiming to be a scikit-learn core contributor, which is the perfect cover for a guy whose only real contribution to machine learning is teaching his algorithm how to get an engagement rate of 1.2 percent.
About François
Into peopleware as much as software, I am passionate about machine learning, open source and innovation.
Highlights
- Top 5% in Computer Software — Ranked #107 of 4683 creators
- Consistent Creator — 22 posts in 30d · top 5%
- Top 5% in France — Ranked #22 of 527 creators
- Top Engager — 1.69% rate · top 25%
Recent posts
The same team is now shaping the abstractions and standards that will underpin agentic machine learning at Probabl. It’s a new paradigm, with many fundamentals still to be reinvented.
2 reactions · 0 comments · 0 reposts
I'm recruiting a post-doc to work on Tabular Foundation Models, one of the hotest topics in AI, where we are at the leading edge. If you're a highly motivated early career researcher, my group, Soda - Inria, is an excellent place for this research and one of the best places to learn about TFMs. To know more about the position please read the full description, and follow the instructions to apply https://lnkd.in/eR8p3p3T
490 reactions · 6 comments · 83 reposts
I had a lot of fun with Maxime Gabella, discussing our dreams about AI. We both see that the next frontier of AI is building agents that use statistical learning to tackle problem-specific challenges from data. And this is where it gets hard. Indeed, the challenge with agentic coding is that it has a tendency to cheat; for instance it is known to modify tests to make them pass. When doing data science, agentic coding falls into traps, but these traps are much harder to catch than a modified test. To have an AI data scientist, there is a need for a statistical harness, that adds statistical e
104 reactions · 10 comments · 12 reposts