Most conversational AI is built for English, leaving 400M+ Arabic speakers underserved. I built an interactive platform that brings Arabic historical figures to life through voice — using a RAG architecture grounded in verified knowledge bases to ensure factual, hallucination-free responses.The system integrates a FAISS vector database with Matryoshka embeddings for semantic search, Arabic TTS/STT via ElevenLabs and Google Web Speech API, and 3D avatars with custom lip-sync algorithms that map Arabic phonemes to realistic facial animations — all connected through a three-tier FastAPI architecture.