Kaggle Learn
by Kaggle
Seventeen hands-on micro-courses plus Google's 5-day Gen AI and agents intensives, all in the browser.
Overview
Kaggle Learn is Kaggle's free micro-course platform, and everything runs inside Kaggle's hosted notebooks, so there is no local environment to configure. The catalogue is a ladder of short courses — Python, Pandas, Data Visualization, Intro to Machine Learning, Intermediate Machine Learning, Data Cleaning, Feature Engineering, Time Series, Geospatial Analysis, Machine Learning Explainability, Intro to Deep Learning, Computer Vision, Natural Language Processing, Intro to Game AI and Reinforcement Learning, Intro to AI Ethics, Intro to SQL and Advanced SQL — each built as a handful of short tutorial pages, every one paired with a graded exercise notebook on a real dataset. Intro to Machine Learning runs from how models work through validation, underfitting and overfitting to random forests and a competition submission; Intermediate Machine Learning covers missing values, categorical encodings, pipelines, cross-validation, XGBoost and data leakage; Feature Engineering covers mutual information, creating features, K-means clustering, PCA and target encoding. Kaggle has since added a second tier, the Learn Guides, built with Google: the 5-Day Gen AI Intensive (foundation models and prompt engineering, embeddings and vector databases, AI agents with LangGraph, domain-specific LLMs such as SecLM and Med-PaLM, and MLOps on Vertex AI) and the June 2026 5-Day AI Agents Intensive, which drew more than 353,000 registrations, 392,000 Discord participants and over 6,000 capstone submissions. Both are now permanently self-paced, combining expert-written whitepapers, NotebookLM audio summaries and Kaggle codelabs, and need only a phone-verified account.
At a Glance
- Topic
- ML
- Level
- All Levels
- Format
- Interactive
- Cost
- Free
- Duration
- Self-paced — each micro-course is a few hours; the 5-day intensives run five days of whitepapers plus codelabs
- Provider
- Kaggle
- Hands-on
- Yes — code/exercises
- Certificate
- Available
What You’ll Learn
- ✓Build and validate a first scikit-learn model, then diagnose over- and underfitting
- ✓Handle missing values, categorical encodings, pipelines and cross-validation properly
- ✓Engineer features using mutual information, K-means clustering, PCA and target encoding
- ✓Avoid data leakage — the failure mode that quietly inflates every score
- ✓Train neural networks and convolutional models for tabular data and computer vision
- ✓Query real datasets in BigQuery with SQL, including analytic and nested queries
- ✓Build agents and RAG pipelines on the Gemini API in the 5-day intensives
Highlights
- •Zero setup — every lesson and exercise runs in Kaggle's hosted notebooks
- •Exercises are graded on real datasets and lead directly into Kaggle competitions
- •The Google-built 5-Day Gen AI and AI Agents intensives are now permanently self-paced
- •The June 2026 AI Agents intensive drew 353,000+ registrations and 6,000+ capstone projects
- •Honest caveat: the micro-courses are deliberately shallow — they teach the API surface, not the mathematics
Who It’s For
Best For
- ✓Beginners who learn by doing rather than by watching lectures
- ✓Analysts adding practical machine learning without a mathematics detour
- ✓Engineers who want a fast, free on-ramp to the Gemini API and agents
- ✓Anyone preparing to enter their first Kaggle competition
Prerequisites
- •None for the Python course — it starts from Hello, Python
- •Basic Python before the machine-learning and deep-learning tracks
- •A phone-verified Kaggle account to run notebooks and open the Learn Guides
FAQ
What is Kaggle Learn?
Kaggle's free, entirely in-browser micro-courses, from Python and pandas through feature engineering, deep learning and SQL — every lesson paired with a graded exercise notebook on a real dataset. The newer Learn Guides add Google's 5-day Gen AI and AI Agents intensives, so the same platform now covers the Gemini API, RAG, agents and MLOps as well as the fundamentals.
Is Kaggle Learn free?
Kaggle Learn is free to access.
What level is Kaggle Learn for?
Kaggle Learn is aimed at a all levels audience. Recommended background: None for the Python course — it starts from Hello, Python, Basic Python before the machine-learning and deep-learning tracks, A phone-verified Kaggle account to run notebooks and open the Learn Guides.
How long does Kaggle Learn take?
Expect roughly Self-paced — each micro-course is a few hours; the 5-day intensives run five days of whitepapers plus codelabs. Most learners work through it at their own pace.
What will I learn from Kaggle Learn?
You'll learn: Build and validate a first scikit-learn model, then diagnose over- and underfitting; Handle missing values, categorical encodings, pipelines and cross-validation properly; Engineer features using mutual information, K-means clustering, PCA and target encoding; Avoid data leakage — the failure mode that quietly inflates every score; Train neural networks and convolutional models for tabular data and computer vision; Query real datasets in BigQuery with SQL, including analytic and nested queries; Build agents and RAG pipelines on the Gemini API in the 5-day intensives.
Topics
Sources
This page was written from 5 sources, 4 on domains other than kaggle.com.