Senior Voice AI Engineer Needed — LiveKit Runtime and Multi-Agent Voice System

Senior Voice AI Engineer Needed — LiveKit Runtime and Multi-Agent Voice System We are looking for a senior engineer to build a production-grade voice AI runtime for enterprise customer-service bots. This is not a basic chatbot or simple STT → LLM → TTS project. The system will use LiveKit Agents and should support: * Real-time voice conversations * Interruptions and barge-in * A fast conversational LLM that speaks with the customer * A second LLM that searches documents and enterprise systems in parallel * Structured customer-service workflows * Tool calling with validation, retries, and idempotency * Policy enforcement before business actions * Response verification before speech is generated * Human handoff * Full tracing, latency, and cost monitoring Core Architecture Phone / Browser ↓ LiveKit Agents ↓ Session Supervisor ├── Conversation LLM ├── Research LLM ├── Workflow Engine ├── Policy Engine ├── Tool Gateway └── Response Verifier ↓ CRM / Billing / Ticketing / Knowledge Base Initial Scope The first version should include: * LiveKit voice session * Streaming STT and TTS * Turn detection and interruptions * Session state management * Parallel research jobs * One structured customer-service workflow * Five to ten mock tools * Tool validation and retry handling * Response verification * Full event tracing * Automated tests Preferred Skills * Python * LiveKit Agents * Asyncio * Real-time voice systems * WebRTC or telephony * LLM tool calling * Multi-agent or supervisor architecture * FastAPI * Redis * PostgreSQL * Temporal * OpenTelemetry Experience with Pipecat, Vapi, Retell, Twilio Media Streams, or native audio models is a plus. Who Should Apply Please apply only if you have built real-time voice or agent infrastructure before. We are not looking for someone whose experience is limited to: * Prompt engineering * Basic chatbots * No-code voice-agent platforms * Simple Twilio integrations * Basic RAG demos When Applying Please include: 1. Relevant voice AI or real-time systems you have built 2. Your experience with LiveKit Agents 3. How you would manage two LLMs working in parallel during a live call 4. How you would prevent duplicate or unsafe tool execution 5. Links to relevant GitHub repositories or demos