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 session live 
 
 $  
 
 
 Question 14 / 35 02 pts, practical 
 

You fit a Pipeline with StandardScaler + LogisticRegression. Which call gives unbiased generalization error ?

 
     
 
 
 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 — Practitioners
 
Individual data scientists
 
Validate your applied scikit-learn skills.
 

Strengthen your profile with a trusted, verifiable credential. Gain confidence for hiring,
          promotion, or freelance work.

 
     
  •  Verifiable digital badge for LinkedIn & CV

    •  
  •  Hands-on assessment, not multiple-choice trivia

    •  
  •  Recognized by hiring managers and consulting firms

    •  
 Get certified 
 
 
02 — Team leads
 
Engineering managers
 
Establish a shared technical baseline.
 

Objectively assess team skill levels. Support career development with a credible standard.
          Reduce subjective hiring decisions.

 
     
  •  Skill-mapped to scikit-learn maintainers’ rubric

    •  
  •  Bulk programs with a single dashboard

    •  
  •  Same exam your candidates take

    •  
 Certify your team 
 
 
03 — Academic
 
Universities & exec ed
 
Industry-recognized capstone for your curricula.
 

Complement curricula with a practical, industry-recognized assessment. Improve graduate
          employability with verified credentials.

 
     
  •  Co-branded with your institution

    •  
  •  Aligns with real-world ML expectations

    •  
  •  Used by CentraleSupélec & Artefact School of Data

    •  
 Integrate it 
 
 
04 — Consulting firms
 
Service providers
 
A clear signal of technical excellence.
 

Certify consultants at scale. Strengthen credibility in proposals and staffing decisions.
          Provide clients with proof.

 
     
  •  Volume pricing for partner programs

    •  
  •  Shared client-facing badge directory

    •  
  •  Same proctoring standard as individuals

    •  
 Join the program 
 
 
 
  
 
 
 Three levels, one rubric 

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

 
 
 
 
LEVEL 01
 
Associate Practitioner
 
Foundations of applied ML with scikit-learn
 
 Junior, 0 to 2 yrs·120 min, 35 questions 
 
 

What we test

 
     
  • Estimators API
    •  
  • Train/test splits
    •  
  • Pipelines & ColumnTransformer
    •  
  • Cross-validation basics
    •  
  • Common metrics
    •  
 Register for Associate 
 
 
 
LEVEL 02
 
Professional
 
Production-grade modelling, hands-on
 
 Mid, 2 to 5 yrs·120 min, 35 questions 
 
 

What we test

 
     
  • Nested cross-validation
    •  
  • Feature engineering with skrub
    •  
  • Calibration & threshold tuning
    •  
  • Imbalanced learning
    •  
  • Model interpretability
    •  
 Register for Professional 
 
 
 
LEVEL 03
 
Expert
 
Advanced internals, research-grade rigor
 
 Senior, 5+ yrs·120 min, 35 questions 
 
 

What we test

 
     
  • Custom estimators (sklearn API)
    •  
  • Fairness & drift monitoring
    •  
  • Causal inference patterns
    •  
  • Probabilistic forecasts
    •  
  • Reproducibility audits
    •  
 Register for Expert 
 
 
 
 
  
 
 
 
 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.
 
 
 
Training
  Where can I find training to prepare for the certification ?  
Three online courses on Skolar. Associate and Professional are live, Expert is in production. Each course aligns with one certification level and mirrors the typical career progression.
 
 
  Is Skolar free to use ?  
Yes, create a free account and access the Practitioner course at no cost. Premium content may follow.
 
 
  Will completing the Practitioner track guarantee I pass ?  
Scoring above 80% on the wrap-up quizzes strongly indicates you are ready to take and pass the corresponding exam.
 
 
Certification
  Do I have to pass all 3 levels in order ?  
No, you can pass any level independently. The 5-minute self-assessment helps you pick the right one.
 
 
  What is the exam format ?  
120 minutes, up to 35 MCQs (60% theory, 40% hands-on). Theory questions are worth 1 to 2 points, practical 2 to 3. You need 70% to pass. JupyterLite and the scikit-learn docs are available during the practical section.
 
 
  Can I reschedule my session ?  
Online: 24 hours notice. Onsite: 72 hours. Late changes may incur an additional fee.
 
 
  Which OS is supported ?  
Mac OS and Windows for online proctoring. Linux candidates can take the exam at a Testing Center, native Linux support is in progress.
 
 
  How many attempts do I have ?  
Two attempts per year. Payment is required for each attempt.
 
  
 
 
  
 
 
 
 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
Professionalissued 2026, ID 7K-2C9F 
 
BY PROBABL