Workflowchatpdf-ai
ChatPDF.ai
A microservices-based RAG chatbot for PDF ingestion, vector storage in Pinecone, and low-latency conversational retrieval with streaming responses.
stack
- OpenAI
- FastAPI
- React
- Pinecone
- Tailwind
- Docker
- TypeScript
Architecture
- PDF ingestion service — text extraction, chunking, OpenAI embeddings, vectors stored in Pinecone
- Semantic chunking — splits on paragraph and section boundaries with 100-token overlap to preserve cross-boundary context
- Retrieval service — embeds the query, runs Pinecone similarity search, returns top-k chunks
- Generation service — retrieved chunks passed as context to OpenAI with the question for grounded answers
- Streaming React frontend — responses streamed token-by-token
Results
- Low-latency retrieval via Pinecone vector similarity search
- Streaming responses for real-time chat UX
- Microservices enable independent scaling of each pipeline stage
Action · Send message
Hiring for a remote AI engineering role?
Contact me