Two ways in.
One destination.
Machine learning sits at the top of two staircases. Climb whichever suits the brain you have today — they meet at the same place.
The Math way
Start from arithmetic. Every lesson has a figure you drag with your own hands and unlimited problems that grade themselves. By the end, backpropagation is just the chain rule — and you already know the chain rule.
Foundations free forever →
Calculus I → Linear Algebra → Calculus II →
Probability → Statistics → Optimization →
Machine Learning: from Math to Models
Start the Math way — free
Calculus I → Linear Algebra → Calculus II →
Probability → Statistics → Optimization →
Machine Learning: from Math to Models
no account · no card · 26 languages
The Python way
Write your first print() today; read real gradient-descent code by lesson 27. No installs, nothing to set up — read-and-predict lessons with auto-graded practice, in your browser.
Python from Zero first lesson free →
variables → loops → lists → functions →
NumPy → plotting →
reading real gradient descent
Start the Python way — free
variables → loops → lists → functions →
NumPy → plotting →
reading real gradient descent
no account · no card · no installs
Or interleave them — one math lesson, one Python lesson.
They're designed to meet at Machine Learning.