Machine Learning Crash Course
by Google
Google's 15-hour ML foundation course, rebuilt in 2024 to cover LLMs and AutoML.
Overview
Google rebuilt this course from the ground up in November 2024, six years after the 2018 original, and the current version is roughly 15 hours of self-study across four sections. ML Models covers linear regression, logistic regression and classification. Data covers working with numerical data, working with categorical data, and datasets, generalization and overfitting. Advanced ML Models covers neural networks, embeddings, and a new Intro to Large Language Models module. Real-world ML covers production ML systems, a new AutoML module, and ML fairness. The LLM module alone runs about 45 minutes and walks through tokens, n-gram models, context, recurrent neural networks, transformers and self-attention, then fine-tuning, distillation and prompt engineering. Teaching is deliberately interactive: animated video explanations, interactive visualizations of algorithms, more than 130 practice questions, and programming exercises that run in Colab with nothing to install locally. Each section ends in a quiz that awards a completion badge, though there is no formal certificate. The prework is spelled out rather than assumed, namely Google's own Introduction to Machine Learning course plus the NumPy and pandas UltraQuick Tutorial Colabs, and the math assumed is algebra, linear algebra as far as tensors and matrix multiplication, and basic statistics, with calculus optional for the advanced modules.
At a Glance
- Topic
- ML
- Level
- Beginner
- Format
- Course
- Cost
- Free
- Duration
- ~15 hours, self-paced
- Provider
- Hands-on
- Yes — code/exercises
- Certificate
- None
What You’ll Learn
- ✓Build and evaluate linear and logistic regression models for classification
- ✓Work with numerical and categorical data as model input features
- ✓Diagnose overfitting using training, validation and test set splits
- ✓Train neural networks and use embeddings to represent sparse data
- ✓Explain how tokens, n-grams, transformers and self-attention power LLMs
- ✓Distinguish fine-tuning, distillation and prompt engineering as LLM adaptation methods
- ✓Apply AutoML and production ML system patterns to real deployments
- ✓Evaluate models for fairness and identify bias in training data
Highlights
- •Completely rebuilt in November 2024 — the 2018 original predated generative AI entirely
- •More than 130 practice questions, with per-section quizzes that award completion badges
- •Every programming exercise runs in Colab, so there is nothing to install or configure
- •Interactive visualizations of algorithms in motion rather than static lecture slides
- •The 2018 launch drew 1,926 points and 222 comments on Hacker News, and it has been a default recommendation since
Who It’s For
Best For
- ✓Developers with Python experience taking their first structured pass at ML
- ✓Engineers who use LLM APIs but never learned the underlying ML fundamentals
- ✓Analysts, PMs and SEO practitioners who need working literacy, not research depth
- ✓Anyone wanting a free, credible refresher before a harder course like CS229
Prerequisites
- •Comfortable programming in Python: functions, dictionaries, lists, loops, list comprehensions
- •The prework: Google's Introduction to Machine Learning course plus the NumPy and pandas UltraQuick Tutorial Colabs
- •Algebra and basic statistics; linear algebra as far as tensors and matrix multiplication
- •Calculus (derivatives, gradients, chain rule) is optional but helps in the advanced modules
FAQ
What is Machine Learning Crash Course?
Google's free, self-paced introduction to machine learning, rebuilt in November 2024 to cover large language models and AutoML alongside the fundamentals. Roughly 15 hours of animated video, interactive visualizations, more than 130 practice questions and Colab exercises take you from linear regression through neural networks, embeddings and production ML, with nothing to install.
Is Machine Learning Crash Course free?
Machine Learning Crash Course is free to access.
What level is Machine Learning Crash Course for?
Machine Learning Crash Course is aimed at a beginner audience. Recommended background: Comfortable programming in Python: functions, dictionaries, lists, loops, list comprehensions, The prework: Google's Introduction to Machine Learning course plus the NumPy and pandas UltraQuick Tutorial Colabs, Algebra and basic statistics; linear algebra as far as tensors and matrix multiplication, Calculus (derivatives, gradients, chain rule) is optional but helps in the advanced modules.
How long does Machine Learning Crash Course take?
Expect roughly ~15 hours, self-paced. Most learners work through it at their own pace.
What will I learn from Machine Learning Crash Course?
You'll learn: Build and evaluate linear and logistic regression models for classification; Work with numerical and categorical data as model input features; Diagnose overfitting using training, validation and test set splits; Train neural networks and use embeddings to represent sparse data; Explain how tokens, n-grams, transformers and self-attention power LLMs; Distinguish fine-tuning, distillation and prompt engineering as LLM adaptation methods; Apply AutoML and production ML system patterns to real deployments; Evaluate models for fairness and identify bias in training data.
Topics
Sources
This page was written from 6 sources, 3 on domains other than developers.google.com.