We're building a voice telephony product for medical practices in Switzerland. Think: AI phone assistant that handles appointment scheduling, prescription requests, and call routing in German and Swiss German. We've done the market validation and have a paying client ready to pilot. Now we need someone who's actually built voice AI pipelines in production. What we're looking for: - Experience with real-time voice agent architecture (STT + LLM + TTS pipeline, or speech-to-speech models) - Hands-on with frameworks like LiveKit Agents, Pipecat, or similar - Familiar with Whisper (ideally fine-tuned variants), streaming TTS (Azure Neural, Cartesia, ElevenLabs), and telephony integration (SIP trunks, Twilio, Telnyx) - Bonus: German/Swiss German language experience, healthcare or regulated industry background About us: We're a small Swiss AI consultancy (two founders + freelancers). We work directly with enterprise clients on AI products. This project has a clear scope and budget, but we have a growing pipeline of similar work. If we click, there's more to come. What this is NOT: - Not a full-time position - Not "build us a chatbot" What we need from you: - Your take on the right architecture for a low-latency, German-speaking voice phone assistant - A rough estimate of what it would take to get to a working MVP - Examples of voice AI projects you've shipped If this sounds interesting, DM me or drop a comment. Happy to share more details about the technical requirements and scope.