Voice AI & Infrastructure Engineer

Neurabit Solution; United States; Remote; USD 110,000 - 165,000; Full time

Job summary

Neurabit Solution is seeking an engineer to own a production voice-agent product from conversation design through supporting services and infrastructure. You will focus on making real-time conversations natural and dependable, then scale the system for high call volumes with low latency.

You will improve conversation flows, diagnose issues, design prompts, and tune the speech pipeline to sub-500ms latency. You will build and operate telephony and cloud services, providing analytics and traces

Qualifications

• 2–4 years of experience and evidence of shipping at least one real-time voice agent into production.
• Strong Python skills and ability to own backend services end to end.
• Hands-on experience with a real-time voice or language API, or a speech-to-text, language-model, and text-to-speech pipeline; practical prompt and flow design skills.
• Experience deploying and operating services on AWS, including containers and queues.
• Strong debugging skills for non-deterministic systems, including tracing the causes of loops or hallucinations.

Responsibilities

• Improve conversation flows, turn-taking, and interruption handling across voice agents.
• Diagnose issues such as mispronunciations, generated speech artifacts, repeated loops, and inconsistent execution; add monitoring to catch regressions.
• Design prompts and flows for real-time language and voice APIs, choosing between prompt changes and architectural fixes.
• Tune the speech pipeline from voice activity detection and transcription through language generation and speech output against a sub-500ms latency target.
• Build and operate telephony, inbound and outbound call handling, and multi-tenant cloud services, including deployment and scaling.
• Provide call analytics, latency traces, and audit trails to make production behavior easier to investigate.

Skills

Python

Real-time systems

Backend development

AWS

Containers

Debugging

Voice API

Prompt/flow design

Tools

MongoDB

LiveKit

WebRTC

Pipecat

Job description

Role overview

Own a production voice-agent product from conversation design through the services and infrastructure that run it. The work focuses on making real-time conversations natural and dependable, then scaling the system for high call volumes with low latency.

Responsibilities

• Improve conversation flows, turn-taking, and interruption handling across voice agents.
• Diagnose issues such as mispronunciations, generated speech artifacts, repeated loops, and inconsistent execution; add monitoring to catch regressions.
• Design prompts and flows for real-time language and voice APIs, choosing between prompt changes and architectural fixes.
• Tune the speech pipeline from voice activity detection and transcription through language generation and speech output against a sub-500ms latency target.
• Build and operate telephony, inbound and outbound call handling, and multi-tenant cloud services, including deployment and scaling.
• Provide call analytics, latency traces, and audit trails to make production behavior easier to investigate.

Requirements

• 2–4 years of experience and evidence of shipping at least one real-time voice agent into production.
• Strong Python skills and ability to own backend services end to end.
• Hands-on experience with a real-time voice or language API, or a speech-to-text, language-model, and text-to-speech pipeline; practical prompt and flow design skills.
• Experience deploying and operating services on AWS, including containers and queues.
• Strong debugging skills for non-deterministic systems, including tracing the causes of loops or hallucinations.

Nice to have

• WebRTC, SIP, telephony, LiveKit, or Pipecat experience.
• MongoDB, event pipelines, analytics, or multi-tenant SaaS scaling experience.
• Hinglish or multilingual conversation design; browser automation or scraping experience.