AI learning platform — solo build
Ilimi
A full-stack AI product built end to end by one person: give it a topic, and it generates a structured course pathway with modules, sections and adaptive quizzes.
What it is
Ilimi turns a topic you want to learn into a persistent, structured course — pathways, modules, sections and adaptive quizzes generated from a single prompt. I built it after noticing that general-purpose AI chat gives useful answers but leaves knowledge scattered across disconnected chat histories.
How it’s built
A nine-model relational schema (pathways → modules → sections, quizzes → questions → options) sits behind end-to-end typed tRPC procedures with Zod validation.
Production concerns are shipped, not sketched: Kinde authentication, Paystack subscriptions with webhook handling, and Sentry monitoring across client, server and edge runtimes.
Highlights
- →Solo full-stack build — 65 commits, zero co-authors
- →Nine-model relational schema, end-to-end typed
- →Auth, subscriptions and monitoring in production