Every one of the 452 modules runs the same loop, and every clip below is the real thing: a real lesson from the Robotics course, recorded section by section. Nothing here is a mock-up.
The seven lettered sections every concept lesson and project shares, plus the quiz, the mastery gate, the build check that closes a project, and the review queue that brings all of it back weeks later.
The idea in plain words, then the exact syntax you will need.
Every lesson opens the same way: what the thing is and why it exists, then a worked example you can run, then a scannable reference block stating the general form the problems later in the lesson will assume. You are never asked to write something the page has not shown you the shape of.
It asks before it tells.
The answer is hidden behind a click. You commit to an answer, reveal it, and rate yourself honestly, and that rating is what moves your mastery score. Reading a worked answer feels like learning; being asked and getting it wrong is learning.
Why, not how.
Short written answers that probe whether you understood the reason for a rule or only memorised it. What you type is saved as you type, so you can come back to your own words weeks later and see whether you still agree with them.
You write the code. Then you compare.
Real problems with a code editor built into the page, and the model solution beside yours once you have committed. The clip shows a genuine near-miss: the bounds tested loosest-first, so one branch can never fire, which is exactly the mistake this problem exists to catch.
Explain it to someone who has never programmed.
The hardest test of understanding is teaching it without jargon. Every lesson ends with one, and the lesson will not close until you have written it.
The mistakes you are about to make, named in advance.
Each one comes with how to recognise it and how to avoid it. Anything you actually get wrong, whether in the quiz, in practice or in a review, is written to your own error log automatically, so your weak points become a list you can work through.
One harder problem, with no scaffolding.
Same material, no worked example, no hints. It is the check that you can use the idea rather than follow a recipe.
Graded on the spot, and what you miss is logged.
Four questions on every module. Your score is pooled with your recall ratings into a single mastery number, and a wrong answer goes straight to your error log so it comes back to you later.
You cannot mark a lesson complete below 80%.
The button stays dead until pooled recall and quiz clear 80%, the practice problems are marked done, and the Feynman explanation is written. There is no skip button. It is the feature that makes every section above it mean something.
A project completes when the build is ticked, not when you say so.
The 144 projects across the four courses each carry a walkthrough of real steps. The step count is read from the module itself, so the gate cannot be talked out of by the browser.
It asks again at 1, 3, 7, 14, 30 and 60 days.
Finishing a lesson schedules it. The review asks real questions rather than showing you the card again, and reschedules each module from measured accuracy. It only ever asks about material you have actually completed.
That is one lesson. Around all 452 of them sits the rest of the system: what you build, what you can practise for free, and what keeps bringing the material back long after you first met it.
Every topic ends in a project you build yourself and push to GitHub, 144 of them across the four courses. Four flagship builds trace the automation-to-robotics path employers actually interview on, so you leave with work, not a certificate.
Every topic has a real-hardware lab, and every lab has a free simulator path first : Gazebo, OpenPLC, KiCad, ngspice. You get the thing working in software before you spend anything on parts, and only buy once you know you will use them.
The answer to every question is hidden until you have tried to produce it yourself. Struggling for a few seconds and getting it wrong is what makes it stick; reading a worked answer only feels like it does.
Finishing a lesson puts it in a queue that brings it back after 1, 3, 7, 14, 30 and 60 days , and the review asks you fresh questions rather than showing you the same card, so you find out what you have actually kept.
A lesson will not let you mark it finished until you have scored 80% or better, worked the problems and written the plain-English explanation. Scrolling to the bottom does not reveal it.
Anything you get wrong is written to your own error log automatically, with how to recognise it next time. Reviews mix topics together rather than drilling one, which is harder in the moment and far more durable.
(x, y) coordinate in robot code?
Decrypt ▸
Four career tracks, each the whole domain, with the same retention system running through every module.