The Absolute Minimum You Must Know

Welcome

For every topic in software development there is a minimum viable set of ideas that unlocks everything else. Each chapter in this book is one page: read it in ten minutes, know enough to work — and, just as importantly in the age of AI, know enough to direct the machine and catch it when it’s wrong.

That second half is the point. An AI assistant will happily write your code, your queries, and your shell commands. What it cannot do is know what you meant, or check its own work with your interests at heart. The minimum you must know is no longer “enough to do everything by hand” — it is enough to decompose a problem, give a precise instruction, and verify the result. Every chapter here ends with exactly that: how to direct an AI at the topic, and a plausible-but-wrong AI answer for you to catch.

NoteThis book checks its own examples

Every Python example in these pages is a real doctest, executed on every change by continuous integration. If an example were wrong, the build would fail. The book practices the verification it preaches.

Three trails through the same chapters

Read cover to cover if you like — the parts are ordered deliberately — or follow the trail that matches why you’re here:

I want to analyse data → Colab · Python · Pandas · Matplotlib · Visualisation · SQLite

I want to build software → Linux · Git · Python · Functions · OOP · Testing

I want to direct AI well → Computational Thinking · Git · Testing · Doctest · AI Coding Agents

I’m brand new to all of this → Start Here · Python · Functions · Linux — and keep the Glossary open in a second tab.

Reading with an AI beside you

These pages use each domain’s real jargon on purpose — the words are part of the minimum, because they’re how you’ll search, read documentation, and direct an AI. When you hit a term you don’t know: first the Glossary; for anything beyond it, ask an AI chat — precisely. The prompt shape that works:

“In the context of Git, what does ‘staging’ mean? I understand files and folders but not version control. Two sentences and a tiny example.”

Name the context, say what you already know, and size the answer. That’s directing the machine applied to your own learning — and the same caution applies: explanations can be confabulated too, so check what the AI tells you against the page in front of you.

Where should I start?

Tick everything you can honestly do right now, without looking anything up. The list of chapters you still need appears below. (If most of these statements don’t even parse yet, that’s fine — skip the quiz and go straight to Start Here.)

I can navigate, read files, and fix permissions in a Linux terminal — and escape vim. I can explain what git add does that git commit doesn't, and safely undo a pushed commit. I can explain why a mutable default argument is a bug. I can say what a function returns when it has no return statement. I know when to reach for a dict or set instead of a list, and why it matters at scale. I can decompose a fuzzy request into precise, ordered steps before writing any code. I can explain when a class earns its keep — and when a function is the better tool. I can write a test that would fail if the code were wrong (not one that always passes). I can turn a usage example into a doctest and run it. I can filter a DataFrame with a boolean mask without googling it. I can explain why SQL queries must be parameterised, and write one that is. I can explain the notebook hidden-state problem and how Restart & Run All exposes it. Before letting an AI agent edit my code, I commit first — and I review its diff after.

Using this book

• On the web — you’re here. Search is in the sidebar; every page has an edit link.
• In a real environment — the repository ships a dev container: open it in GitHub Codespaces and you get a genuine Linux shell, Git, and Python with every library these pages use — nothing to install, free tier, one click.
• As plain files — every chapter is a standalone Markdown file readable straight on GitHub. git clone and read in your editor.

That’s the tour. Pick a trail — or tick the boxes above and let the book pick for you.