How to use this course without wasting your time

TL;DR: Read one lesson, do its exercise, take its drill, then move on. Do not read ahead and do not read the concept pages first. The lessons carry the path, the concept pages are the reference you consult when a lesson points at one, and the question bank is where you find out whether any of it stuck.

Where you are. Last lesson of the orientation module. The next module opens the machinery, and this one is about how to travel through it. It is the shortest lesson in the course and the one most worth not skipping, because the three failure modes below are common enough that avoiding them is worth more than any single lesson.

Three ways people waste weeks

Reading without doing. By far the most common. Every lesson ends with a ten-minute exercise, and doing them is roughly the difference between finishing this course able to build something and finishing it able to talk about building something. The exercises are deliberately small: count tokens, compare distances, ask a question with a false premise. None takes an evening, and each produces a number or a result you did not already know.

Reading everything before building anything. Seven modules is a lot of material, and the instinct is to get through it and then start. That order wastes most of the value, because the material only fixes itself against real problems. If you have a project, start it after module 4 and let the rest of the course answer questions you have actually hit.

Chasing the ecosystem. Frameworks and tools change constantly, and it is easy to spend a month learning one and have it be irrelevant. This course teaches mechanisms rather than products for that reason: what any retrieval system must solve, what any agent framework must handle. Learn the mechanism and you can pick up any tool in an afternoon. Learn the tool and you have learned one tool.

How this fits with the rest of the site

Three surfaces, three jobs, and using them in the wrong order is the second failure mode in disguise.

Lessons carry the narrative: what to learn, in what order, and why it matters. Read these in sequence.

Concept pages are the reference for a single idea in depth. Every lesson links to a few. Read one when a lesson sends you there or when you want the mechanism it deliberately skipped. Do not read the curriculum cover to cover before starting; 214 concept pages read in isolation is exactly the "reading without doing" failure at scale.

Questions are how you find out whether it stuck. Each lesson links to a few real interview questions on its topic. Attempt them yourself before reading the answer, even roughly, even out loud. A question you have tried and failed teaches more than three you have read.

What is free and what is not

Modules 1 to 4 are free to read in full, and they are the majority of the technical foundation. Later modules ask you to sign in or subscribe, and the question bank has its own free allowance.

The thing to know: every lesson is complete on its own. Nothing here withholds the point of a lesson to sell you the next one. Where a lesson links to something paid, it is because the deep answer to that particular question genuinely lives there, and the lesson still stands without it.

A realistic pace

Seven modules, roughly forty lessons, ten to fifteen minutes each plus an exercise. That is around fifteen hours of reading and perhaps the same again of doing.

At a lesson a day it is six weeks. At three a week it is three months. Both work. What does not work is eight lessons in one sitting, because the exercises get skipped and by the fourth lesson nothing is landing.

If you are preparing for an interview with a deadline, do modules 1 to 4 in order, then jump to module 8, then come back. If you are learning the craft with no deadline, go in order and take the projects.

Do this before moving on

Decide two things and write them down.

First, your pace: how many lessons a week, on which days. Vague intentions do not survive a busy fortnight.

Second, what you will build. Any small thing that uses one of the four shapes from the first lesson: a tool that extracts fields from documents you already have, or answers questions from notes you already keep. Having a real project makes every later lesson concrete, and you will be able to start it properly after module 4.

Go deeper

• The concept curriculum is the reference layer described above. Browse the structure now so you know what is there, then leave it until a lesson sends you.
• The essentials is the curated set of must-know interview questions, and it is the right entry point if your goal is an interview rather than the craft.
• Tokenization is a good first concept page to read, because the next lesson leans on it and it is short.
• Practice question: How is AI engineering different from ML engineering? is worth attempting now as a baseline, so you can compare your answer with the one you would give after module 7.
• Practice question: When do you build an agent instead of a single LLM call? will be beyond you today, and that is the point. Note where it loses you and revisit after module 5.

Key takeaways

• One lesson, its exercise, its drill, then move on. The exercises are where the learning actually happens.
• Do not read the concept curriculum first. It is a reference, consulted when a lesson points at it.
• Learn mechanisms rather than tools; a tool takes an afternoon once you know what problem it solves.
• Start building after module 4 and let the rest of the course answer questions you have actually hit.

LEARNING LAB1 of 4

Check yourself before an interviewer does. Answer from memory first.

What is the most common way people waste time on material like this?

Reading the lessons out of orderReading without doing the exercises, which is the difference between being able to build and being able to talk about buildingNot reading the concept pages thoroughly enough first