Remote Senior Voice AI Engineer - Real-Time Latency Expert

Clera; United States; Remote; USD 96,000 - 110,000; Full time

Job summary

Clera is seeking a founding engineer to own the real-time voice layer, from speech input through AI reasoning to spoken output. You will help ensure end-to-end latency is minimized in production, delivering natural, responsive voice interactions.

You will build streaming speech-to-text, text-to-speech, and transport layers while collaborating with the founders on architectural decisions in a fast-moving, fully remote team with core overlap 13:00–17:00 UTC.

Qualifications

• 5+ years building production software.
• 2+ years shipping voice/real-time audio.
• Strong Python or TypeScript skills.
• Experience with audio stacks such as LiveKit, Pipecat, Vapi, Twilio Media Streams, Daily.
• Experience debugging audio at frame level (sample rates, codecs, jitter).
• Experience building LLM evaluation harnesses and optimizing latency.

Responsibilities

• Build and own streaming speech-to-text, LLM turn-taking, text-to-speech, and transport.
• Measure and reduce latency, targeting first audio under 800 ms on real calls.
• Address interruptions, barge-in, silence detection, overlapping speech, accents.
• Build an evaluation harness from recordings to detect regressions.
• Compare voice providers and models through evidence-based testing.
• Instrument production systems for turn latency and cost per minute.
• Work directly with founders and make technical decisions in a fast-moving team.

Skills

Real-time audio

Latency optimization

Python/TypeScript

Speech tech

Tools

LiveKit

Twilio Media Streams

WebRTC

Pipecat

Vapi

WebSocket

Daily

Job description

Clera is seeking a founding engineer to own the real-time voice layer, from speech input through AI reasoning to spoken output. You will help ensure end-to-end latency is minimized in production, delivering natural, responsive voice interactions.

You will build streaming speech-to-text, text-to-speech, and transport layers while collaborating with the founders on architectural decisions in a fast-moving, fully remote team with core overlap 13:00–17:00 UTC.