Search Jobs

Search by job, company or skills

Voice AI Engineer (WebRTC / LiveKit)

Voice AI Engineer (WebRTC / LiveKit)

neurodrift
1-4 Years
Early Applicant
  • Posted 23 hours ago
  • Be among the first 10 applicants

Job Description

Voice AI Engineer (WebRTC / LiveKit)

NeuroDrift is a US-based voice AI company. Our agents run on real phone lines every day for enterprise contact centres, so latency, telephony quirks and audio edge cases are our daily work.

We run LiveKit self-hosted: the server, the SIP service, Egress and Agents workers on Kubernetes. We are hiring an engineer who has run this stack, not one who has only built agents on top of it.

What you'll do

- ⁠ ⁠Own the LiveKit stack in production: server, SIP service, Egress, Agents workers

- ⁠ ⁠Connect and harden SIP trunks against real carriers and contact-centre platforms: transfers, headers, call correlation

- ⁠ ⁠Run stateful agent workers: capacity, scaling, graceful drain, deploys during live calls

- ⁠ ⁠Own latency from end-of-utterance to first audio, and know which component owns each millisecond

- ⁠ ⁠Build the observability that makes a bad call explainable after the fact

- ⁠ ⁠Work with the agent team on STT/TTS/LLM, turn-taking and interruption behaviour

What we need

- ⁠ ⁠1-2+ years Python, asyncio

- ⁠ ⁠Self-hosted LiveKit server in production: config, TURN, Redis multi-node, upgrades

- ⁠ ⁠LiveKit SIP service on a carrier trunk: inbound/outbound trunks, dispatch rules, transfers

- ⁠ ⁠LiveKit Agents workers in production: dispatch, scaling, graceful drain

- ⁠ ⁠WebRTC, SIP, RTP

- ⁠ ⁠Docker; Kubernetes preferred

- ⁠ ⁠Cross-service call debugging: SIP, server and agent logs

Nice to have

- ⁠ ⁠Voice agent build: STT/TTS/LLM pipeline, tool calling, VAD, interruption tuning

- ⁠ ⁠LiveKit Egress

- ⁠ ⁠SIP trunk debugged under load: Telnyx, Twilio, Bandwidth, Vonage

- ⁠ ⁠Latency instrumentation: end-of-utterance, time-to-first-audio, p50/p90

- ⁠ ⁠CCaaS integration: Genesys, Five9, NICE, SIP REFER, UUI headers

- ⁠ ⁠Other media servers: Janus, mediasoup, FreeSWITCH, Asterisk, Kamailio

- ⁠ ⁠STT/TTS per-minute cost awareness

The setup

- ⁠ ⁠Fully remote

- ⁠ ⁠High intensity, 50 to 60 hours a week. Startup pace, not a 9-to-5

- ⁠ ⁠Working hours primarily IST, with availability for US client meetings

If your LiveKit experience is LiveKit Cloud plus the Agents quickstart, this is not the role.

Key Skills

LiveKit server

CCaaS integration

Latency instrumentation

Call debugging

Egress

Agents workers

SIP service

STT

TURN

Python asyncio

Dispatch rules

About Company