I need a backend developer to build the first version of an AI-powered voice receptionist for a medical clinic in Germany. The system must handle incoming phone calls through Twilio, stream the caller’s audio in real time to an AI model (Open AI Realtime or equivalent), receive the AI’s audio/text response, and play it back to the caller with low latency. The agent should understand what the caller wants (appointments, rescheduling, questions, administrative issues) and use simple backend “tools” to perform actions such as creating an appointment request or creating a staff task. All code must be in English, while the AI’s spoken conversation will be in German (you don’t need to speak German). The logic of the agent must not be hard-coded; clinical content, FAQs, and dialogue rules should be configurable through external JSON/DB so I can edit them without touching code. The final deliverable is a Node.js/TypeScript service with Twilio webhooks, a media-stream → AI loop, basic conversation state management, logging to PostgreSQL, and deployment to my Azure App Service. I’m looking for a clean, well-structured MVP that I can easily extend later for more clinics.