Building Lumora – Our Hackathon Story

What Inspired Us

We built Lumora during one of the busiest, most chaotic weeks of our lives.

Sisa, a trainee accountant, was deep in prep for his mock exams on his journey to becoming a Chartered Accountant. Tiisetso was juggling multiple intense work projects—including another hackathon at the same time.

Despite all this, we both felt strongly about building something meaningful. As former university roommates, we often looked back at how overwhelming and fragmented our learning paths were—especially in Computer Science (for Tiisetso) and Accounting Science (for Sisa). We always thought, “What if there was a way to actually learn smarter, not just harder?” That idea became Lumora.

What We Built

Lumora is an AI-powered learning platform that takes your goals, available time, and preferences—and turns them into a structured, personalized curriculum you can actually follow.

We built:

A full onboarding flow with Supabase auth A powerful AI-backed curriculum generator with dynamic pacing A beautiful UI with module timelines, expandables, quizzes, and tracked progress Chat-based AI assistant embedded into every plan Responsive layout with animations and skeletons for fast UX Everything was styled with shadcn/ui, built in Next.js, and wired with Supabase as our backend.

What We Learned

This project pushed us to level up in:

Prompt engineering: It took many iterations to craft the perfect system prompt for real, valuable AI-generated learning plans. Collaboration under pressure: Our schedules were hectic, our syncs were short—but we learned how to communicate clearly and divide tasks quickly. GitHub version control: We sharpened our workflow with proper branches, commits, and PRs after a few chaotic merges early on. Resilience and adaptability: Between long load-shedding power cuts and missed meetings, we had to learn how to just keep moving.

Challenges We Faced

Power outages: We lost electricity for 48 hours right before final submission. That meant late-night coding sprints and occasional hotspotting from our phones just to push a build. Late ideation: It took us forever just to lock in the core concept. We had to move fast and cut ruthlessly once we had it. Time pressure: With real work deadlines and exams looming, every hour counted. We had to build smart and lean. Prompt tuning: Getting our AI to consistently return structured, deep educational content—not just placeholder modules—was a huge challenge we’re proud we cracked.

Why We’re Proud

We didn’t just build a hackathon project. We built something we wish we had in university—and something we still plan to use today.

And more than that, we built it together. Even in the chaos, we shipped.

What’s Next

We're already working on:

Multi-skill plans (e.g. “Frontend + UX”) Real-time plan export and sharing A community of AI-powered learners who learn together Thanks for following our journey. — Sisa & Tiisetso 🧠⚡

Built With

  • bolt
  • supabase
  • vite
Share this project:

Updates