Document Intelligence SaaS: LangGraph Agent RAG for PDF Search

This project involved building a multi-tenant document intelligence SaaS that turns large volumes of PDFs into searchable, answerable knowledge.

The core is a LangGraph RAG pipeline in which agents choose the right tools for parsing, retrieval and answering.

OpenAI embeddings are stored in Weaviate for semantic search, while document metadata is stored in PostgreSQL.

The backend was built as FastAPI microservices that handle document ingestion and context-aware responses.

The final platform lets organizations upload large PDF collections and get relevant, context-aware answers from them within a multi-tenant SaaS architecture.