The challenge
Most online learning platforms feel slow, impersonal, and rely on passive video content. Windy combines self-paced, project-based courses with AI-powered tutoring and real-time feedback so learners build momentum through doing—not just watching.
Where I focused
Defined product vision, learning model, and full technical architecture—from Rails + React island hybrid to AI tutor integration, gamification rules, and monetization strategy.
How we approached it
Learning engine design
Architected course hierarchy (Course → Module → Lesson → Project Checkpoint → Capstone) with block-level progress persistence and rubric-based AI feedback on project submissions.
Hybrid frontend architecture
Chose Rails 8 + Hotwire for fast server-rendered pages (catalog, auth, billing) with React islands for high-interaction surfaces (lesson player, quiz engine, AI tutor chat, milestone celebrations).
AI tutor integration
Designed context-aware tutoring system scoped to lesson and project context, with session history, safety filters, and provider abstraction for LLM flexibility.
Gamification system
Built bounded gamification with points, badges, and daily streaks tied to meaningful milestones—first course started, first project completed, first quiz passed—without competitive leaderboards.
What shipped
- Complete product and technical spec covering 7 application domains and 17 API endpoints.
- Architecture validated: Rails + Hotwire for 80% of surfaces, React islands for lesson player, quiz engine, and AI chat.
- Performance targets defined: <50ms interaction feedback, <180ms API read P95, 60fps animations.
- Monetization model finalized: hybrid subscription tiers with one-time premium course purchases via Stripe.
What I learned
Great learning products feel fast and personal. Windy's hybrid architecture proves you can deliver premium interactive experiences where they matter most while keeping the rest of the product simple, fast, and maintainable.
