RAG-Chat
RAG-Chat is a retrieval-augmented generation application built around asynchronous document ingestion, OpenAI embeddings, Qdrant vector search, BullMQ workers, and citation-backed responses.
RAG-Chat was built to understand the full document-to-answer pipeline in a retrieval-augmented system instead of stopping at a basic chat UI. Users can upload files, process them in the background, store embeddings in a vector index, and ask questions that return grounded answers tied back to source chunks.
RAG-Chat was built to understand the full document-to-answer pipeline in a retrieval-augmented system instead of stopping at a basic chat UI. Users can upload files, process them in the background, store embeddings in a vector index, and ask questions that return grounded answers tied back to source chunks.
The architecture separates ingestion from serving: upload routes enqueue work into BullMQ, background workers extract and chunk text with format-specific parsers, embeddings are stored in Qdrant, and chat routes retrieve relevant context before generating responses. The result is a more realistic document workflow with processing status, semantic search, and citation-backed answers.