RecallAI — AI-Powered Flashcard Learning System

---

✨ RecallAI transforms your study materials into interactive flashcard decks instantly

Powered by advanced AI models • SM-2 Spaced Repetition • Real-time Analytics

---

🚀 Quick Start

git clone https://github.com/kinshukkush/recallai.git
cd recallai
npm install
npm run dev

Open http://localhost:3000 in your browser.

---

🎯 Core Features

Feature	Description
📄 PDF Upload & Parsing	Drag and drop any PDF — textbooks, notes, papers with automatic text extraction
🤖 AI Flashcard Generation	Advanced AI creates targeted Q&A with dynamic difficulty levels
🔄 Active Recall Practice	Animated 3D flip cards test your knowledge one question at a time
⚡ Spaced Repetition (SM-2)	Proven algorithm schedules reviews at optimal intervals
📊 Progress Dashboard	Track mastery, weak spots, and cards due today with beautiful analytics
💡 "Explain Better" AI Mode	Ask AI to simplify complex answers with plain language

---

🏗️ Architecture

┌─────────────────────────────────────────────────────────┐
│          RecallAI Processing Pipeline                   │
├─────────────────────────────────────────────────────────┤
│                                                         │
│  [PDF Upload] → [Parse & Extract] → [AI Generation]     │
│                                         ↓               │
│                                  [Supabase DB]          │
│                                         ↓               │
│          [Review Practice] ← [SM-2 Algorithm]           │
│                                                         │
└─────────────────────────────────────────────────────────┘

---

🛠️ Tech Stack

---

🤖 AI Integration

Primary Model: Google Gemma-4 (via Ollama)

RecallAI leverages Ollama with Gemma-4 for local, privacy-first AI processing:

Why Ollama + Gemma-4?

• 🔐 Privacy: All processing happens locally on your machine
• ⚡ Performance: Lightning-fast inference with optimized models
• 💰 Cost-effective: No expensive API calls
• 🎯 Specialized: Fine-tuned for reasoning and code generation

Setup Instructions

1. Install Ollama from ollama.ai

2. Pull the Gemma-4 model:

   ollama pull gemma4

3. Alternative models available:

   ollama pull gemma4:26b      # Mixture of Experts (4B active)
   ollama pull gemma4:e4b      # Efficient 4B variant
   ollama pull gemma4:e2b      # Efficient 2B variant (mobile)

4. Run with Claude Code (optional IDE integration):

   ollama launch claude --model gemma4

5. Environment Configuration:

   # .env.local
   OLLAMA_API_URL=http://localhost:11434
   OLLAMA_MODEL=gemma4

Model Specifications

Model	Size	Context	Best For
gemma4:latest	9.6GB	128K	Balanced performance
gemma4:26b	18GB	256K	High-quality output
gemma4:31b	20GB	256K	Frontier performance
gemma4:e4b	7.2GB	128K	Efficient inference
gemma4:e2b	5.1GB	128K	Mobile/Edge devices

Key Capabilities

✅ Reasoning - Multi-step problem solving
✅ Coding - Generate and explain code
✅ Vision - Image understanding (supported models)
✅ Audio - Audio processing (supported models)
✅ Tools - Native function-calling support
✅ Native System Prompts - Structured conversations

Performance Benchmarks

Benchmark	Gemma-4 31B	Gemma-4 E4B
MMLU Pro	85.2%	69.4%
AIME 2026	89.2%	42.5%
LiveCodeBench	80.0%	52.0%
Codeforces ELO	2150	940

---

📦 API Endpoints

Endpoint	Method	Purpose
/api/generate-cards	POST	Process PDF and generate AI flashcards
/api/review-card	POST	Submit review and apply SM-2 algorithm
/api/explain	POST	Simplify complex flashcard answers
/api/decks	GET	Retrieve all user decks
/api/decks	DELETE	Delete deck and associated cards
/api/decks/[id]	GET	Fetch specific deck with cards

---

🔧 Configuration & Setup

Prerequisites

• Node.js 18+ & npm
• Ollama (for local AI processing)
• Supabase account

Environment Variables

Create a .env.local file:

# Ollama Configuration
OLLAMA_API_URL=http://localhost:11434
OLLAMA_MODEL=gemma4

# Supabase
NEXT_PUBLIC_SUPABASE_URL=your_supabase_url
NEXT_PUBLIC_SUPABASE_PUBLISHABLE_KEY=your_anon_key
SUPABASE_DB_PASSWORD=your_db_password

Database Setup

1. Create a Supabase project
2. Run the schema from supabase/schema.sql
3. Configure RLS policies (included in schema)

Installation Steps

# 1. Clone repository
git clone https://github.com/kinshukkush/recallai.git
cd recallai

# 2. Install dependencies
npm install

# 3. Start Ollama (in separate terminal)
ollama serve

# 4. Pull Gemma model
ollama pull gemma4

# 5. Configure environment variables
cp .env.example .env.local
# Edit .env.local with your settings

# 6. Start development server
npm run dev

---

📊 Data Flow

User PDF Upload
       ↓
   PDF Parse
       ↓
   Extract Text
       ↓
  Ollama/Gemma-4 AI
       ↓
Generate Q&A Cards
       ↓
Store in Supabase
       ↓
  User Reviews
       ↓
   SM-2 Algorithm
       ↓
Update Intervals

---

🎨 UI/UX Features

• ✨ 3D Card Animations - Smooth flip and tilt effects
• 🌌 Glassmorphism Design - Modern frosted glass aesthetic
• 🎯 Responsive Layouts - Mobile-first design
• ⚡ Framer Motion - Smooth entrance and hover animations
• 🌈 Gradient Accents - Eye-catching color transitions

---

🚀 Performance Optimizations

• Image Optimization - Next.js automatic image compression
• Code Splitting - Dynamic imports for faster load times
• Caching Strategy - Efficient cache headers
• Database Indexing - Optimized Supabase queries

---

🔐 Security

• ✅ Row-Level Security (RLS) in Supabase
• ✅ Environment variable protection
• ✅ Secure API route handling
• ✅ Input validation and sanitization

---

📈 Learning Analytics

RecallAI provides comprehensive insights:

• Mastery Percentage - Overall progress tracking
• Due Cards - Cards scheduled for today
• Learning Streak - Consistent practice motivation
• Difficulty Breakdown - Identify weak areas
• Time Analytics - Study duration trends

---

🎓 How It Works

1. Upload - Drag and drop your PDF
2. Parse - RecallAI extracts and cleans text
3. Generate - Gemma-4 AI creates smart questions
4. Review - Practice with animated flashcards
5. Learn - SM-2 adapts to your pace

---

🤝 Contributing

Contributions are welcome! Please follow our code style and submit PRs with clear descriptions.

# Fork the repo, then:
git checkout -b feature/your-feature
git commit -m "Add your feature"
git push origin feature/your-feature

---

📝 License

MIT License - see LICENSE file for details

---

👨‍💻 Developer

Kinshuk Saxena

Made with ❤️ and passion for education by Kinshuk Saxena

⭐ Star this repo if you found it helpful! ⭐