📄 RAG-Based PDF QA System

This project provides a REST API built with FastAPI that enables question answering over PDF documents using LangChain. It allows users to ask questions, which are answered by a Large Language Model (LLM) based on the information contained in a predefined PDF document.

🚀 Features

• ✅ Load and split content from a PDF file
• ✅ Lexical retrieval using BM25 and TF-IDF
• ✅ Re-ranking with Flashrank
• ✅ Use of EnsembleRetriever + ContextualCompressionRetriever
• ✅ RAG pipeline powered by LangChain
• ✅ HTTP endpoint built with FastAPI
• ✅ Configuration via .env file

🧠 Technologies

• Python 3.10+
• LangChain
• FastAPI
• OpenAI / LM Studio
• BM25 & TF-IDF retriever
• PyPDFLoader

Installation

git clone https://github.com/kullaniciadi/rag-pdf-qa.git
cd rag-pdf-qa```

python -m venv venv
source venv/bin/activate

pip install -r requirements.txt

Environment Variables

To run this project, you will need to add the following environment variables to your .env file

PDF_PATH LLM_API_KEY LLM_API_BASE RERANKER_MODEL EMBEDDING_MODEL

Usage/Examples

uvicorn main:app --reload