We are building a production-grade real-time AI Voice Agent and need a hands-on LiveKit expert to help us architect, debug, and optimize our full call pipeline. You'll work directly with our engineering team to reduce latency, improve audio quality, and make the voice experience feel natural and human. Our core pipeline runs: STT → RAG → LLM → TTS, delivered over WebRTC/PSTN via LiveKit. What you'll work on: - Diagnose and reduce end-to-end latency across all pipeline stages - Eliminate robotic audio artifacts, jitter, and packet loss - Optimize WebRTC/SIP media routing in LiveKit (RTP, trunking, agent sessions) - Tune Deepgram and Cartesia for speed and audio naturalness - Improve Pinecone retrieval performance and caching - Ensure clean interruption handling and conversational turn-taking - Provide architecture guidance and code review as needed