Valentina14142000/cross-lingual-rag-engine

A high-performance RAG middleware designed to slash LLM latency and compute costs by caching semantically equivalent queries across multiple languages (English, Swahili, Russian, etc.). # 🌐 Cross-Lingual Semantic Cache & Hybrid RAG Engine A high-performance RAG middleware designed to slash LLM latency and compute costs by caching semantically equivalent queries across multiple languages (**English, Swahili, Russian, etc.**). --- ## 🏗️ Architecture * **Semantic Cache:** Uses multilingual sentence embeddings (`paraphrase-multilingual-MiniLM-L12-v2`) and cosine similarity to serve instant cache hits for paraphrased or cross-lingual queries. * **Hybrid Retrieval:** Falls back to a combined dense vector and sparse keyword (BM25) search pipeline on cache misses. --- ## 🚀 Quick Start ### 1. Setup Environment ```bash python3 -m venv venv source venv/bin/activate pip install sentence-transformers rank_bm25 numpy fastapi uvicorn pydantic ``` ### 2. Run Server ```Bash uvicorn main:app --reload --port 8001 ``` ### 3. Test API ```Bash curl -X POST 127.0.0.1 \ -H "Content-Type: application/json" \ -d '{"query": "How do high-throughput AI systems use Rust and PyTorch?"}' ```