A learning coach for AI.
Pick a topic. Take a 2-minute skill check. Get a short list of free and paid courses actually matched to where you are, not where some algorithm assumes you are.
Live: https://cyborg1244.github.io/uncoded/
The most valuable skill of the next decade is the ability to learn, unlearn, and relearn quickly. The internet has more learning material than ever, and most of it is the wrong fit for the person reading it. This is a small tool to fix that for AI specifically.
- Choose what you want to learn (LLMs, ML foundations, MLOps, prompt engineering, etc.).
- Answer ~5 quick questions that gauge where you actually are.
- Get a curated, ranked shortlist of free and paid courses that match your topic and level, with a one-line note on why each one fits.
No tracking. No accounts. No hallucinated course links. Every recommendation is hand-curated.
- Vite + React 19 + TypeScript
- Plain CSS (no UI framework)
- Static — deployable anywhere. Hosted on GitHub Pages via Actions.
npm install
npm run devFound a great course that should be here? Or a course on the list that's gone stale? Open an issue or a PR against src/data/courses.ts.
Course entries follow this shape:
{
title: "...",
provider: "...",
url: "https://...",
cost: "free" | "paid" | "freemium",
durationHours: 12,
level: "beginner" | "intermediate" | "advanced",
topics: ["llms-agents", "..."],
why: "One sentence on why this course is worth your time at this level.",
}The bar: it has to be something you'd recommend to a friend without caveats.
- More topics (RL, multi-modal, evals, AI safety)
- Optional LLM layer that personalizes the framing of recommendations (never invents the courses themselves)
- Save your level for a topic and check back in 6 months for a follow-up plan
- "What I built" — a way for users to share what they shipped after taking a course
MIT