💻 More and more people are talking about agents for #datascience. Fewer people talk about what makes them reliable.
Here are 6 terms every data scientist should know, handpicked by our CPO @YannDebray based on the evidence we’ve found while building Skore 🧵
1
48
➡️ Guardrails: leakage and baseline checks, then a human approves the export.
📊 Want to know more about these concepts ? Pre-register to receive the full Sate of Tabular AI Report:
2decv6.share-eu1.hsforms.com…
13
⚠️ Spoiler alert: The State of Tabular AI Report 2026 lands on Nov 24!
Written by @GaelVaroquaux and Cailean Osborne with a foreword by Madelon Hulsebos 👀
⚒️ Python data science libraries are the foundational tools of Tabular AI, sustained by global communities committed to keeping them open and free.
🚀 Tabular foundation models are a fast-moving frontier, propelled by a culture of open science shared among the leading labs.
1
46
🤖 Agentic data science is on the rise! Learn about how agents accelerate workflows and shift the bottleneck from pipeline building to validating agent-driven experiments.
🫖 Sounds like your cup of tea? Pre-register to get it in you inbox! 2decv6.share-eu1.hsforms.com…
30
Data scientists are increasingly using copilots and coding agents to accelerate their workflows, so we asked @thomasjpfan what role @scikit_learn plays now.
His answer:
“Generative AI does not replace scikit-learn. It uses it.”
1
143
As Thomas put it:
“When a workflow needs a random forest, a coding agent does not need to waste tokens writing a performant random forest from scratch.”
This is thanks to the work he has spent years doing alongside his fellow maintainers and contributors.
🌏 Scikit-learn is built by thousands of people around the world. Every month, we feature one of them. Who should we talk to next?
Read the full interview here👇
probabl.ai/blog/community-vo…
29
🏎️ @scikit_learn 1.9 deep dive: Did you know you can accelerate your scikit-learn workflows on #GPUs?
From 82 seconds to 2 seconds. See how Emily Chen reduced training time of a ridge regression model.
▶️ Watch the full deep dive: youtu.be/j-Iq7iXQvEA
💡The #Python array API standard offers a solution. It is a shared specification for NumPy-like operations, adopted by the major array libraries.








