epythonlab2/legal-advisor

AI-powered legal question assistant for Ethiopian laws using Large Language Models (LLMs) and Retrieval Augmented Generation (RAG). # Ethio Legal Assistant AI-powered legal question assistant for Ethiopian laws using Large Language Models (LLMs) and Retrieval Augmented Generation (RAG). ## Project Vision Legal information in Ethiopia is often difficult to access, scattered across PDFs, and hard to search.\ This project builds a local AI assistant that allows users to ask legal questions in natural language and receive answers grounded in Ethiopian legal documents. Example questions: - What are employee rights under Ethiopian labor law? - What is the legal process for land disputes? - What penalties exist for contract violations? ------------------------------------------------------------------------ ## Features - Natural language legal Q&A - Retrieval from Ethiopian legal documents - Local LLM inference (CPU friendly) - Source citations from legal texts - Modular architecture Planned: - Amharic language support - Voice interface - Legal document summarization - Case law search ------------------------------------------------------------------------ ## Architecture User Question\ ↓\ Embedding Model\ ↓\ Vector Database Search\ ↓\ Relevant Legal Documents\ ↓\ LLM Generates Answer\ ↓\ Response with Sources Main components: 1. Document processing pipeline 2. Vector database 3. Embedding model 4. Local LLM 5. API / Interface ------------------------------------------------------------------------ ## Technology Stack - Python - LangChain - FAISS Vector Database - SentenceTransformers - Local LLM (Mistral / Llama / Phi) - FastAPI - Streamlit Optional tools: - Ollama - HuggingFace Transformers - ChromaDB ------------------------------------------------------------------------ ## Project Structure ethio-legal-assistant │ ├── data │ ├── raw_legal_docs │ └── processed_docs │ ├── embeddings │ └── build_vector_db.py │ ├── rag │ ├── retriever.py │ ├── prompt_template.py │ └── pipeline.py │ ├── models │ └── load_llm.py │ ├── …