Live AI TV Host Developer

Looking for an experienced developer to build a LIVE interactive AI television host. This is NOT a pre-recorded AI avatar/video project. The system must work in real time during an actual live TV broadcast. The TV show already exists and currently uses a human presenter. vMix is already installed and used for the live broadcast, so the developer does not need to build the broadcasting system itself. The AI avatar/audio output needs to be integrated into our existing vMix workflow. Required live pipeline: Viewer phone call → Estonian Speech-to-Text → OpenAI → Estonian Text-to-Speech → HeyGen LiveAvatar → vMix → Live TV Main requirements: • HeyGen LiveAvatar API integration • One realistic avatar for the first version • Incoming viewer telephone calls – Twilio, Vapi or another reliable and cost-effective solution • High-quality Estonian speech recognition (STT) • OpenAI for live conversation and response generation • Conversation context/memory during the call • Natural Estonian TTS voice – ElevenLabs or another suitable solution • Real-time avatar lip-sync and natural movement • Video/audio output integrated with our existing vMix setup • Approximately 3–5 seconds end-to-end response latency is acceptable • System must be reliable enough for live television • Low ongoing API/operating costs are important Important: I am not looking for someone who only creates videos with HeyGen. I need a developer who understands real-time APIs, streaming audio, AI voice agents and live avatar integration. When applying, please answer these questions: Have you previously worked with HeyGen LiveAvatar API? If yes, please provide an example or demo. Have you built a real-time AI voice/phone agent before? How would you technically build this pipeline? What technology would you use to send the LiveAvatar video/audio into vMix? What end-to-end latency do you realistically expect? What is your estimated development time? What is your estimated development cost? What would you estimate the ongoing operating/API cost per hour to be? I want to start with one avatar and a working MVP, test it in our real broadcast environment, and expand the system later if the first version works reliably. Please do not send a generic AI-development proposal. I am specifically interested in developers with relevant real-time voice, streaming, WebRTC/LiveKit and/or LiveAvatar experience.