Ten subjects, in order, from your first line of Python to a monitored system in production. Free, no sign-up, openly licensed.
Nothing is published yet. All ten subjects are in production and each is released only once it is complete, rather than as a trickle of half-finished lessons. Subjects 01 and 02 are being recorded first. The full plan is below so you can see exactly what is coming — and tell us if something important is missing. Get notified when the first subject opens.
New to all of it? Start at subject 01. Already write code? Take the placement quiz — it works now — and skip ahead.
Watch, read the written guide, run the code, then build the project. The project is the part that makes it stick.
Post the question in the community. Someone who was stuck on it last month will answer.
The curriculum
Each subject is a full course: video lessons, a written guide, runnable code, and a project you finish. Filter by stage if you already know where you belong.
The language everything else is written in. Install it, break it, fix it, and learn to read an error message instead of fearing it.
Only the mathematics that actually appears inside models, taught where the need arises — plus an explicit list of what you can safely skip.
The classical toolkit, which still solves most real problems better and cheaper than a neural network does.
Backpropagation by hand before any framework, then the architecture behind every model you have heard of.
Images as numbers, and classifiers that survive photographs taken by an actual phone in actual light.
How machines handle text, and what breaks on low-resource languages — which is most of the world's languages, including Afghanistan's.
Getting genuinely useful work out of large models — and measuring whether the output is right rather than merely fluent.
Systems that decide what to do next, use tools, and hand back to a human when they are unsure.
The part most courses skip, and the part that decides whether your work survives contact with real users.
One complete system, start to finish, on a problem that matters. The portfolio piece an employer will actually read.
Nothing in that stage yet. Try another filter, or tell us what is missing.
Video lessons
Every lesson will be free on YouTube, downloadable for offline viewing, and available at 360p for slow connections.
Subject 01
Subject 02
Subject 04
Subject 07
AI tools directory
You do not need to wait for us to start using these. Everything here has a free tier that is genuinely usable — though limits change often, so check before you rely on one.
| Tool | Use it for | Cost | Note |
|---|---|---|---|
| Google Colab | Running Python and training small models in the browser | Free tier | Start here. No installation, and it gives you a GPU for short sessions. |
| VS Code | Writing and debugging code | Free | Works on a modest laptop. Install the Python extension. |
| Git & GitHub | Version control and showing your work | Free | Your GitHub profile is your portfolio. Employers look at it. |
| Hugging Face | Downloading models and datasets | Free tier | Search for your own language and you will find how little exists. |
| Kaggle | Datasets, notebooks and competitions | Free | Free GPU hours each week, and thousands of notebooks to read. |
| Ollama | Running language models on your own machine | Free | Works offline once downloaded. Needs about 8 GB of memory. |
| scikit-learn | Classical machine learning | Free, open source | Subject 03 lives here. |
| PyTorch | Deep learning | Free, open source | Used from subject 04 onwards. |
| Chroma or FAISS | Vector search for retrieval systems | Free, open source | Either works for subject 07. Chroma is easier to begin with. |
| FastAPI & Docker | Serving and shipping a model | Free, open source | The backbone of subject 09. |
| ChatGPT / Claude / Gemini | Explaining errors, reviewing your code, drafting | Free tiers | Ask it to explain, not to finish your work for you. |
Programming resources
These are free, excellent, and available today. When a tutorial and the official documentation disagree, the documentation is right.
Practical projects
Each brief states the data you need, the subject it belongs to, and what finished means. Do not wait for the course — if you can already code, start today.
Collect wheat or fuel prices for a season, chart them, and forecast next month. Teaches pandas, plotting, and honest uncertainty.
A vocabulary app with spaced repetition, running fully offline. Teaches data structures and shipping something usable.
Three-day particulate forecasts from public sensor and weather data, with a plain-language health warning.
Photograph 300 healthy and diseased leaves, train a classifier, run it on a phone. Teaches small-data reality.
Answer questions from the school curriculum with a page citation, and refuse when the books do not cover it.
An agent that reads a referral, extracts the fields, and asks a human when it is unsure. Teaches safe hand-off.
Placement quiz
This works right now. No score is stored or sent anywhere. Answer honestly — starting too high is the most common reason people give up.
1 / 5