Empowering Indian developers with intelligent, RAG-powered resource discovery
An AI-first marketplace that transforms how developers discover, evaluate, and integrate APIs, models, and datasets through context-aware search with multilingual support for English, Hindi, and Hinglish.
Frontend: Next.js 16 (App Router) • TypeScript • Tailwind CSS • NextAuth
Backend: FastAPI • Python 3.11 • Pydantic • Mangum (Lambda)
AI/ML: AWS Bedrock (Claude 3.5 Sonnet) • Titan Embeddings v2
Database: Neon (Serverless PostgreSQL) • Pinecone (Serverless Vector Store)
Infrastructure: AWS Amplify • EC2 • Redis • S3
Advanced search combining SQL metadata filtering (category, pricing, popularity) with vector similarity ranking. SQL filters narrow the dataset, then vector search ranks by semantic relevance.
Context-aware AI assistant using AWS Bedrock Claude 3.5 Sonnet that understands developer intent and provides personalized resource recommendations with code examples.
Automated data pipeline ingesting from:
- GitHub: 654+ repositories and developer tools
- HuggingFace: 1,100+ ML models and datasets
- Kaggle: 40+ structured datasets
- OpenRouter: 346+ LLM models with pricing
Custom ranking algorithm normalizing scores across different platforms using composite metrics: relevance, popularity, optimization, and freshness with unified rank_score computation.
Native support for English, Hindi, and Hinglish queries with intelligent language detection and culturally-aware response matching for India's 5M+ developers.
Serverless-First Approach with hybrid deployment strategy:
graph TB
A[Next.js 16 Frontend<br/>AWS Amplify] --> B[FastAPI Gateway<br/>EC2 + Gunicorn]
B --> C[Neon PostgreSQL<br/>Serverless Metadata]
B --> D[Pinecone Vector DB<br/>Serverless Search]
B --> E[AWS Bedrock<br/>Claude 3.5 + Titan]
B --> F[Redis Cache<br/>ElastiCache]
G[Ingestion Pipeline<br/>Python Orchestrator] --> C
G --> D
G --> E
Key Design Decisions:
- Frontend: Amplify for global CDN and auto-scaling
- Backend: EC2 for cost optimization with reserved instances
- Database: Neon for serverless PostgreSQL with connection pooling
- Vector Store: Pinecone for sub-100ms semantic search
- AI: Bedrock for enterprise-grade LLM access
- Cache: Redis for distributed locks and embedding cache
dev-store-ai-bharat/
├── backend/ # FastAPI backend
│ ├── main.py # Application entry point
│ ├── api_gateway.py # EC2 deployment gateway
│ ├── config.py # Configuration management
│ ├── clients/ # AWS Bedrock & Database clients
│ ├── routers/ # API route handlers
│ │ ├── search.py # Dual-layer search endpoint
│ │ ├── rag.py # RAG chat assistant
│ │ ├── resources.py # Resource management
│ │ └── auth.py # Authentication
│ ├── services/ # Business logic
│ │ ├── search.py # Search orchestration
│ │ ├── embeddings.py # Vector operations
│ │ └── ranking.py # Score normalization
│ ├── models/ # Pydantic data models
│ ├── ingestion/ # Multi-source pipeline
│ │ ├── orchestrator_production.py # Main pipeline
│ │ ├── fetchers/ # Source-specific harvesters
│ │ ├── services/ # Chunking, embedding, ranking
│ │ ├── stages/ # Atomic pipeline stages
│ │ └── INGESTION_GUIDE.md # Pipeline documentation
│ ├── migrations/ # Database schema
│ └── tests/ # Test suite
├── frontend/ # Next.js 16 frontend
│ ├── app/ # App Router pages
│ ├── components/ # React components
│ │ └── DevStoreDashboard.jsx # Main dashboard
│ ├── lib/ # API client utilities
│ └── auth.ts # NextAuth configuration
├── docs/ # Documentation
└── .kiro/specs/ # Feature specifications
git clone https://github.com/your-org/dev-store-ai-bharat.git
cd dev-store-ai-bharat# Backend
cp backend/.env.example backend/.env
# Frontend
cp frontend/.env.local.example frontend/.env.local
# Configure your API keys (see .env files for details)# Terminal 1 - Backend
cd backend && pip install -r requirements.txt
uvicorn main:app --reload --port 8000
# Terminal 2 - Frontend
cd frontend && npm install && npm run dev
# Or use the convenience script
./start_all.sh🌐 Access: Frontend at http://localhost:3000 • API docs at http://localhost:8000/docs
- Frontend: AWS Amplify with global CDN (configured via
amplify.yml) - Backend: EC2 with Application Load Balancer (see
backend/deploy.sh) - Database: Neon PostgreSQL with connection pooling
- Monitoring: Custom health checks and systemd services
# Frontend (Amplify auto-deploys on push)
git push origin main
# Backend (EC2 deployment)
cd backend && ./deploy.sh- Search Latency: <200ms average response time
- Vector Search: Sub-100ms with Pinecone serverless optimization
- Pipeline Throughput: 10-15 resources/sec (Bedrock embedding limited)
- Full Sync: 3-5 minutes for 2,500+ resources
- Scale: Optimized for 100K+ resource tier
| Endpoint | Method | Description |
|---|---|---|
/api/v1/search |
POST | Dual-layer semantic search (SQL + Vector) |
/api/v1/rag/chat |
POST | RAG-powered AI assistant |
/api/v1/resources |
GET | Resource listing with filters |
/api/v1/resources/{id} |
GET | Resource details with metadata |
/api/v1/health |
GET | System health and metrics |
Run the automated multi-source ingestion:
cd backend
# Production pipeline (requires AWS credentials)
python ingestion/run_production.py
# Development pipeline (JSON output only)
python ingestion/run_ingestion.pyPipeline Stages: Fetch → Normalize → Deduplicate → Upsert (Neon) → Embed (Bedrock) → Upsert (Pinecone) → Rank → Cache
See backend/ingestion/INGESTION_GUIDE.md for detailed documentation.
Team Dev-Store-AI-Bharat:
- Mohd Arsh - AI/ML Engineering (RAG, Embeddings, Multilingual)
- Raunak - Data Engineering (Pipelines, Infrastructure, Monitoring)
- Vansh - Frontend Engineering (UI/UX, Performance, Accessibility)
- Aryan - Backend Engineering (APIs, DevOps, Architecture)
- 🔧 Backend Setup Guide
- 🤖 RAG System Documentation
- 📊 Ranking Algorithm Details
- 🔄 Ingestion Pipeline Guide
- 🚀 EC2 Deployment Guide
Apache License 2.0 - Built with ❤️ for the Indian developer community