skii-bee/SA-RAG-System

Local-first RAG system for South African documents. Runs completely offline with Ollama + Streamlit. # 🇿🇦 SA Local RAG System **A completely offline, private document Q&A system that runs on a laptop with only 8GB RAM.** Built with Python, LangChain, Ollama, and Streamlit. ## 🔍 What This Does Upload PDFs, Word documents, or text files. Ask questions in natural language. Get answers based **only** on your documents. No data ever leaves your machine. ## 🛠️ Tech Stack - **Local LLM:** Ollama + Llama 3.2 (1B/3B) - **Embeddings:** Fastembed / TF-IDF - **Vector Search:** FAISS / Custom similarity - **UI:** Streamlit - **Language:** Python ## 🚀 Why I Built This I completed my first year of an IT degree coding entirely from my phone. When I finally got a laptop, I wanted to build something that proved I could ship working AI systems under real constraints. This project runs on **8GB RAM** and requires no internet after setup. That's not a limitation – it's a design requirement for real-world deployment. ## 📋 Features - Upload multiple document types (PDF, DOCX, TXT) - Automatic text chunking and embedding - Local LLM inference (no API calls) - Source attribution for answers - Persistent storage of indexed documents ## 🔧 Installation ```bash # Clone the repository git clone github.com cd sa-rag-system # Create virtual environment python -m venv venv source venv/bin/activate # On Windows: venv\Scripts\activate # Install dependencies pip install -r requirements.txt # Pull Ollama models ollama pull llama3.2:1b ollama pull nomic-embed-text # Run the app streamlit run app.py ``` sa-rag-system/ ├── app.py # Streamlit UI ├── document_processor.py # PDF/DOCX loading & chunking ├── vector_store_simple.py # Embeddings & similarity search ├── requirements.txt # Dependencies └── README.md # This file Save the file. --- ## Step 6: Commit and Push Now push everything to GitHub: ```bash # Add README git add README.md # Commit with a meaningful message git commit -m "Initial commit: SA Local …