Agent Platform Engineer, Voice

New York

Development - Back End

Permanent

Our client, an early-stage AI Startup, is hiring an Agent Platform Engineer to join their team in New York City. The successful candidate will lead the development of production voice agents, create reliable, natural customer interactions and build shared tools, memory and infrastructure to support agents across the wider platform.

Responsibilities

Develop and maintain production voice agents that handle inbound and outbound customer interactions.
Refine conversational performance, including instruction adherence, turn-taking, interruptions, pacing, pronunciation and audio quality.
Optimize response times across telephony, speech recognition, model processing, tool execution and speech generation while preserving accuracy and reliability.
Build and manage infrastructure for call routing, queues, transfers, recordings, webhooks and failover.
Create reusable agent capabilities for shared context, instructions, tools, memory, policies, identity and action execution.
Connect agents with scheduling, diagnostic, CRM and operational platforms, ensuring reliable handling of retries, partial failures and ongoing workflows.
Develop automated evaluations, simulations and regression tests to assess task completion, policy compliance, latency, audio quality and escalation to human teams.
Establish monitoring and diagnostics using conversation traces, transcripts, audio events, failure analysis and alerts.
Work closely with product and customer teams to assess real conversations, identify issues and implement improvements.
Help define the architecture and development roadmap for the company’s unified agent platform.

Skillset

At least 2 years of experience in professional software engineering.
A track record of building and deploying backend, real-time, telephony or conversational systems in production.
Strong coding ability in TypeScript or Python.
Sound knowledge of asynchronous processing, streaming, distributed systems and observability.
The ability to balance low latency with reliable execution, instruction adherence and natural conversation.
Confident debugging skills across application code, model responses, audio pipelines and customer configurations.
An interest in developing reusable agent functionality that works across different channels.
A proactive mindset, clear communication skills and the ability to take ownership in a fast-paced startup.
Experience with Twilio, SIP or contact-centre infrastructure is desirable.
Familiarity with LLM agents, streaming speech recognition, text-to-speech, real-time model APIs, evaluation frameworks or human handoff workflows.