aceL - AI Learning Platform

Full-stack adaptive learning web app powered by Featherless AI. Students master any topic through personalised AI tutoring, adaptive learning paths, and collaborative learning.

Features

• Multimodal input: Typed notes, PDF text, handwritten images (OCR), YouTube links. One unified checklist from all materials.
• Learning style: 5-question onboarding → Visual, Reading/Writing, Auditory/Video, or Kinesthetic mode. Content and quiz style adapt.
• Core flow: Upload → Checklist → Feynman/Quiz assessment → Case 1 (mastery) or Case 2 (learning path). Re-assessment until mastery.
• At-risk support: AI Personal Tutor chat (scaffolded, one question at a time), alternative methods, learning style switch.
• Collaborative (opt-in): Study groups, peer explanation review, community analogies (when implemented with Supabase).

Tech stack

• Frontend: Next.js 14 (App Router), Tailwind CSS
• AI: Featherless AI for chat and vision (OCR), using the OpenAI SDK against a configurable base URL
• Auth: NextAuth (email + Google)
• Storage: Supabase (optional; app works with localStorage for demo)
• YouTube: youtube-transcript for captions
• Charts: Recharts (ready for confidence/progress charts)

Setup

1. Clone and install

   cd HackTheEast
   npm install

2. Environment

   Copy .env.example to .env and set the variables there (overview below):

   Featherless AI

   • FEATHERLESS_API_KEY
   • NEXT_PUBLIC_FEATHERLESS_BASE_URL — same purpose as in .env.example; if omitted, the app falls back to https://api.featherless.ai/v1 (see src/lib/llm.ts)

   NextAuth

   • NEXTAUTH_URL — e.g. http://localhost:3000 in development
   • NEXTAUTH_SECRET — e.g. openssl rand -base64 32
   • NEXT_PUBLIC_BASE_URL — origin where the site is served (dev: http://localhost:3000; production: your HTTPS URL)

   Google sign-in

   • GOOGLE_CLIENT_ID, GOOGLE_CLIENT_SECRET — required for Continue with Google (see commented steps and redirect URI ${NEXT_PUBLIC_BASE_URL}/api/auth/callback/google in .env.example)

   Supabase (optional; persistence vs localStorage)

   • NEXT_PUBLIC_SUPABASE_URL, NEXT_PUBLIC_SUPABASE_ANON_KEY, SUPABASE_SERVICE_ROLE_KEY

3. Supabase (optional)

   Run supabase/schema.sql in the Supabase SQL Editor to create base tables. Then run supabase/migrations/001_extended_schema.sql for full persistence (topics, assessments, progress, misconceptions, etc.). If you skip this, the app stores topics and progress in localStorage.

4. Run

   npm run dev

   Open your configured base URL (for local dev, http://localhost:3000). Sign in with “Continue with demo” or Google, then go to Upload to add content.

MVP build order (implemented)

1. Upload (PDF + text) → AI checklist and summary
2. Quiz and explanation mode → score and feedback
3. Case 1 vs Case 2 branching and learning path
4. Learning style profile and adaptive content delivery
5. At-risk detection and alternative method suggestions (logic + tutor)
6. AI Personal Tutor chat with scaffolded conversation
7. Handwritten note OCR and YouTube video summarisation
8. Knowledge map (data ready; full viz can be added)
9. Confidence check-in and gamification (schema/APIs ready)
10. Collaborative features (schema ready; Study Groups, Peer Review, Analogies)
11. Spaced repetition (schema ready; notifications)

Design

• Calm, minimal UI; mobile-friendly.
• Encouraging, non-judgmental AI tone.
• At-risk: “This concept is being a bit stubborn — let’s try a different approach.”