Questions people ask
before starting.

Yes. All 90 days, free, no card. You sign in so your progress survives a cleared browser and follows you between devices. Ninety days is far too long to trust local storage.

The same way you passed exams you had not studied for. Not by being brilliant at 2am, but because a fixed scope and a real deadline beat an open-ended year every single time. Most people who plan to learn this over a year quit in month two, because nothing ever forces the next step. This is 214 hours of work with the syllabus already chosen and the order already argued. The honest catch is that crammed knowledge fades, so every concept here is scheduled to come back days and then weeks later until it holds without notes. You are never trying to hold 90 days in your head. Only today.

A little. It's built for someone around 0.5 to 1.5 out of 5. You've seen some programming, you have real gaps, and nobody taught you the foundations properly. It does not assume a CS degree or any AI experience.

Each day ships a Master Prompt tuned for that exact topic and your exact level. You paste it into Claude or ChatGPT and it teaches you, quizzing you before it explains. Underneath sit three to six hand-picked videos, docs, and articles covering the same ground, which you can read instead, or use to check what you were told.

It will, eventually. Models are fluent whether or not they are right, and on a topic you met an hour ago you cannot yet tell the difference. That is the reason every day also ships three to six resources from people who do this for a living, each one opened and timed by hand: official docs, standards, and teaching from names that are checkable. The prompt is the tutor. Those are the reference you hold it against, and the habit of checking a confident answer against a source is most of what this whole path is trying to build.

Your finish date moves by one day. Ninety days means ninety days from the day you start, so the date it ends on is real and the site shows it to you. Miss three days and you finish three days later, still having done ninety days.

That's the design constraint, and it's checked automatically. Every day carries a time budget and the site refuses to build if a day falls outside 120 to 180 minutes. Days with extra depth put it in an optional Extra Time section that nothing later depends on.

Not primarily, but interview thinking is threaded through it: DSA in small doses across all 90 days, and system design questions asked the way an interviewer would. The goal is to actually be a good engineer; passing the interview is a side effect.

No, and that's deliberate. This is AI Engineering: LLM APIs, structured outputs, retrieval, agents, MCP, evals, and guardrails. No model training, no calculus, no gradient descent, no data science pipelines.

Machine learning, in full: no model training, calculus, or gradient descent. This is AI engineering, which is a different job. Dynamic programming is also left out on purpose, because it costs weeks and transfers to almost nothing outside an interview room. Everything else that matters is in: Python for the first twenty-two days, then JavaScript and TypeScript, data structures including graphs and heaps, SQL and schema design, production engineering, system design, and AI engineering through to agents and evals.

Yes. Every one of the 90 days is a public page: the topic, what you will be able to do by the end of it, why it matters, the concepts it covers, and how it connects to the days either side. Signing in reveals the Master Prompt, the curated resources, the exercise and the recall questions, and turns on progress and mastery tracking. It is free either way.