Upload documents. Ask questions. Get AI-powered answers.
A full-stack document intelligence platform that transforms static documents into interactive knowledge bases using Retrieval-Augmented Generation (RAG). Users upload PDFs and text files, then query them conversationally through a natural language chat interface powered by OpenAI's GPT-4.
Key Features
- Intelligent Document Processing: Automatic text extraction, chunking, and semantic embedding generation
- Conversational AI Interface: Natural language queries with context-aware responses
- Event-Driven Architecture: Asynchronous processing pipeline using AWS EventBridge and SQS
- Vector Search: Fast semantic search powered by Pinecone embeddings
- Secure Authentication: User management and JWT validation via Clerk
- Real-time Chat: Persistent conversation history with document context
Tech Highlights
Frontend: Next.js 14, shadcn/ui, TypeScript
Backend: Go (REST API), Python (FastAPI, async workers)
AI/ML: OpenAI GPT-4.1-mini, text-embedding-3-small
Infrastructure: Docker, LocalStack (S3, SQS, EventBridge), Pinecone, NeonDB
Architecture
Event-driven microservices architecture with async document processing:
- Go API handles uploads and generates presigned S3 URLs
- EventBridge triggers SQS queue on document upload
- Python worker processes documents, generates embeddings, stores in Pinecone
- RAG service retrieves relevant chunks and generates contextual answers using GPT-5
