The official scikit-learn certification, proctored by Probabl

Validate your ML expertise. Officially.

The only certification designed by the people who maintain scikit-learn. Train with Skolar. Get assessed on real ML work. Earn a verifiable credential.

For practitioners For organizations 5-min self-assessment
120 min proctoredJupyterLite includedPass at 70%
~/skolar, proctored sessionlive
$ skolar status
  ✓ scikit-learn 1.7.2 detected
  ✓ JupyterLite kernel ready
  ✓ Proctor handshake established
$
LEVEL 2, PROFESSIONAL REAL ML WORK
200M+scikit-learn downloads / month
3certification levels
120minutes per exam
70%score to pass
Who this is for

Stay close to how best practices are defined, debated, and stress-tested.

    01 / 04
    Three levels, one rubric

    From first fit to production. Pick the level that matches where you are today.

    What we test

      example_question.md

      JupyterLite, scikit-learn 1.7.x
      Train where you will be tested

      Skolar is the same standard you will be evaluated on.

      Three online courses, one per certification level. Practice realistic ML workflows. Strengthen weak spots before the exam.

      Get trained on Skolar
      L01Available

      Associate Practitioner course

      8 weeks, self-paced·14 modules
      Start course
      L02Available

      Professional course

      10 weeks, self-paced·18 modules
      Start course
      L03Coming soon

      Expert course

      12 weeks, self-paced·22 modules
      Get notified
      Exam format

      120 minutes. Real code. Real proctors.

      Theoretical questions are weighted at 1 to 2 points. Practical questions at 2 to 3. JupyterLite and the scikit-learn docs are accessible via embedded links during the practical section.

      Length120 min
      QuestionsUp to 35 MCQ
      Mix60% theory, 40% practical
      Pass mark70%
      ToolingJupyterLite + sklearn docs
      ProctoringOnline or onsite
      5 minutes, free

      How well do you really know scikit-learn ?

      Take Probabl’s official self-assessment. We will point you at the right level, and what to learn next on Skolar.

      Start the assessment Compare levels
      Probabl self-assessment
      FREE, 5 MIN
      Q1

      How do you typically validate a classifier ?

      e.g. K-fold cross-validation, stratified, with appropriate metric
      Q2

      What’s the most sophisticated thing you have shipped ?

      e.g. A boosted-tree pipeline with hyper-param search and drift checks
      Q3

      How comfortable are you reading the scikit-learn source ?

      e.g. I read source when I hit edge cases, I open PRs occasionally
      WE RECOMMENDyour right level
      Open form ↗
      What candidates say

      Built on scikit-learn. Trusted by the people doing the hiring.

      « Took the Level 2 exam today and I really liked it. This is the level I had in mind as a hiring threshold for a data science position. »

      Frits Hermans
      Data Scientist, ING

      « Working hand in hand with the scikit-learn team is a strong recognition of the quality of our data science teams. »

      Vincent Luciani
      CEO, Artefact

      « Participants now obtain, in addition to our training certificate, an official credential validating their mastery of scikit-learn. »

      Frédéric Pascal
      Director, CentraleSupélec

      Used by hiring teams & curricula at

      ING
      CentraleSupélec
      Artefact
      neue fische
      datarockstars
      EDF
      Capgemini
      Inria
      Frequently asked

      Everything else, asked plainly.

      Ready when you are

      Train. Get assessed. Earn the credential · with the people who maintain scikit-learn.

      One verifiable digital badge. Three levels. Every question vetted by a maintainer.

      I’m a practitioner I’m an organization
      VERIFIED
      LEVEL 02scikit-learn
      Professional      issued 2026, ID 7K-2C9F
      BY PROBABL