Data science and AI, one clear picture at a time.
Free, open courses from school maths to research-level AI. Every idea is drawn before it is derived. Every derivation is written out in full. Every algorithm is built from scratch before a library is used.
73
courses in 10 connected tracks
671
modules planned, about 2,684 lessons
160
lessons live so far, with 40 interactives
3+
purpose-made figures in every lesson
Free
to read, for good. No sign-up, no tracking
Try a lesson now
40 modules are live so far. Here are three to start with, one from each part of the curriculum. Each is four lessons of about 25 minutes, with an interactive to play with.
Linear algebra · L1 Core
What a matrix really does: it moves the whole plane and keeps the grid a grid.
Machine learning · L1 Core
How a model learns: read the slope of the loss, take a step downhill, repeat.
Reinforcement learning · L2 Intermediate
The one idea under all of reinforcement learning: value now is reward plus value next.
How every lesson works
Every lesson follows the same path, so once you have read one you know how to read them all.
A question first
One concrete question that the lesson will answer by the end.
Then a picture
The idea is drawn before any equation appears, and the parts are named before the process.
No skipped steps
Every derivation shows every line. “It can be shown” is not allowed.
Real numbers
A small example worked by hand, traced through a figure.
Code from scratch
You build the method in plain Python and NumPy, then meet the library version.
Test yourself
Recall questions and graded exercises with full solutions. Questions come back in later lessons.
Beginners and experts read the same page: full proofs and research notes sit in blocks you can open when you want them. The details are in the style guide.
One connected path
Most people learn this subject by stitching together a maths book, a machine learning course, a deep learning site and a reinforcement learning text, each with its own notation. Here every lesson declares what it depends on, one notation runs through everything, and the tracks feed each other.
12 courses · L0 to L4
→
8 courses · L0 to L3
→
S · Statistics and Data Analysis
8 courses · L1 to L4
→
7 courses · L1 to L4
→
10 courses · L2 to L4
→
7 courses · L2 to L4
→
9 courses · L2 to L4
→
5 courses · L2 to L4
Two tracks run alongside the others from early on:
4 courses · L1 to L4
C · Capstones and Research Practice
3 courses · L1 to L4
Browse all 73 courses, or read the roadmap to see what ships when.
Why build this
No existing resource is deep, visual and complete at the same time. The most visual sites cover a few dozen topics. The most complete textbooks carry about one figure for every page or two. And visual teaching material is costly to keep alive, so some of the best of it is no longer maintained.
Applied AI Mastery is built to close those three gaps:
- Coverage. Maths, statistics, programming, machine learning, deep learning, reinforcement learning, applied AI and engineering under one roof and one notation.
- Pictures. At least three purpose-made figures in every 25-minute lesson, and an interactive in every module. Figures are code, so they can be fixed, themed and translated.
- Upkeep. Every formula is tested by a machine on every change, every source is opened and checked, and every lesson shows when it was last verified.
Where it stands
This is a build in progress, and it is honest about it. 160 lessons are live, and they are marked in review until independent readers have checked them. The full curriculum ships in four rings, each useful on its own.
You can help it get there:
- Learn from it and tell us where a step lost you.
- Review a lesson. A second pair of eyes on the maths is the most valuable gift there is.
- Share it with one person who is trying to learn this.
- Follow along: watch the repository or subscribe to the news feed.