PDF-RAG with DeepSeek LLM

A lightweight PDF Question Answering app built using DeepSeek LLM and LangChain with Streamlit frontend.

🔍 Overview

This project allows users to upload a PDF document and ask natural language questions about its content. The app processes the PDF, splits the content into chunks, embeds them into a vector database using FAISS and DeepSeek embeddings, and retrieves the most relevant chunks to generate answers using DeepSeek LLM.

🚀 Features

• Upload any PDF
• Chunk and embed content using LangChain
• Semantic search using FAISS
• LLM-based question answering using DeepSeek
• Streamlit-based intuitive UI

📂 Project Structure

RAG_WITH_DEEPSEEK/
│
├── deepseek/
│   ├── vector_databases.py      # Loads, chunks, embeds PDF content
│   ├── Rag_pipeline.py          # Retrieval + QA logic
│   ├── frontend.py              # Streamlit app
│   └── requirements.txt
│
├── pdfs/                        # PDF upload folder
├── vectorstore/                 # FAISS vector database folder
├── data.pdf                     # Sample PDF
├── .env                         # Environment variables
└── README.md

⚙️ How It Works

1. Upload a PDF using the Streamlit UI.
2. PDF is parsed using PDFPlumberLoader.
3. Content is chunked via RecursiveCharacterTextSplitter.
4. Chunks are embedded using DeepSeek embeddings via OllamaEmbeddings.
5. Chunks are stored in a FAISS vector store.
6. On a user query, relevant chunks are retrieved and passed to DeepSeek LLM for answering.

🧰 Installation

git clone https://github.com/your-username/pdf-rag-deepseek.git
cd pdf-rag-deepseek
python -m venv deepseek
source deepseek/bin/activate  # or .\deepseek\Scripts\activate on Windows
pip install -r requirements.txt

🏃 Usage

streamlit run frontend.py

📝 Requirements

Check requirements.txt for full list.

🏷️ Tags

RAG, LangChain, DeepSeek, Streamlit, Vector Database, LLM, PDF, FAISS

📄 License

MIT License