I need an AI-driven phone agent, built in Python, that can pick up incoming calls and behave just like a bilingual (Arabic / English) CRM operator. Core knowledge sources • A RAG Vector DB that already holds most of our FAQs and product data. • Our own local CRM tables plus a PostgreSQL instance for live customer, ticket, and order data. Runtime expectations The agent has to listen, understand, and reply in real time through Twilio voice. While speaking, it should be able to: • Query either knowledge source automatically and merge the answers into one coherent response. • Take a new order from the caller, validate it, then commit it into the Order DB. • Detect when the caller needs a form completed, open the web form programmatically, fill the fields, submit, and confirm submission on the same call. • Switch smoothly between male and female voices in both supported languages. Phone-specific features I require • Call forwarding • Voice recognition • Voicemail management Tech stack & approach Async Python, FastAPI (or similar), WebSockets for live event streaming, Twilio Programmable Voice/Chat for bidirectional audio, and any LLM orchestration layer you prefer for Retrieval-Augmented Generation. If you have a proven method for prompt engineering, speaker diarization, or latency reduction, highlight it—every millisecond counts on a live call. I am only interested in freelancers who have shipped something similar: an autonomous, production-ready phone agent that talks to relational databases and external forms. When you reply, link one relevant project and outline your proposed toolchain; that is how I will shortlist candidates.