- Delhi, India
- 3–5 years
Lead the end-to-end voice AI product, owning latency, performance, and reliability of the cascaded pipeline from STT to LLM to TTS.
Responsibilities
- Own the end-to-end latency budget and target sub-1-second response time
- Diagnose and fix cross-cutting bottlenecks across STT, LLM, and TTS legs
- Evaluate and integrate STT/LLM/TTS providers, including in-region India options
- Implement turn detection, interruption, barge-in handling, filler audio, and retry logic
- Instrument and trace the full call lifecycle for diagnosable latency regressions