← Back to Work
Screenshot 1 of 2
AI / RAG
RAG Studio — Document Intelligence
⚡ End-to-end RAG pipeline — document ingestion, embeddings, Pinecone retrieval, and cited answers
About this project
Full-stack document Q&A system for PDF, TXT, Markdown, and DOCX files. The ingestion pipeline parses and chunks documents, creates 384-dimensional Hugging Face embeddings, stores them in a serverless Pinecone vector database, retrieves the most relevant context with cosine similarity, and uses Groq-hosted Llama 3.3 to return grounded answers with source citations.