Software Engineer (Harmony)

monday.com. Our agents hold real phone conversations for businesses, and every one of those calls runs through systems we design and build ourselves: real-time audio pipelines, agent orchestration, telephony integrations, and the tools our customers use to build and monitor their agents.

's backing. Engineers here own features end to end, from design through production, and work shoulder to shoulder with AI researchers, turning models into products.

About monday.com

Harmony is the AI voice venture built inside

We also work at the frontier of voice AI. We're building evaluation methods and benchmarks for voice agents, measuring what actually makes a phone conversation succeed, and we train and fine-tune our own models alongside the best available ones.

We're a small, senior team with startup speed and monday.com

Responsibilities

• Backend services that power live voice conversations: sub-second turn-taking, handling interruptions, thousands of concurrent streams, and staying reliable under load.
• The platform customers use to create, test, and analyze their voice agents.
• Integrations with telephony providers, CRMs, calendars, and the wider monday.com platform.
• Closing the loop between production and research: turning real calls into evals, benchmarks, and training signal for our models.
• Listening to real calls and talking to customers to find where agents fail, then fixing it.
• Observability and performance of everything you ship.

Requirements

• 6+ years building and operating production backend systems.
• Strong TypeScript and Node.js, or deep experience in a comparable stack and the will to switch.
• Experience with high-throughput, stateful, or real-time services in production.
• Solid grounding in APIs, databases, and event-driven architectures.
• A product mindset: you care how the agent sounds to the person on the line, not just how the service performs.
• Experience taking products from 0 to 1 in fast-moving, loosely defined environments.
• Curiosity about AI and ML. Hands-on ML experience is a plus, not a requirement.
• Pragmatic: you ship fast and stay rigorous where it matters.

Preferred

• Real-time audio or streaming (WebRTC, media pipelines), or telephony (SIP).
• STT/TTS, voice AI frameworks (e.g. LiveKit, Pipecat), or evals for LLM-powered systems.
• Don't check every box? Apply anyway. Strong candidates rarely match a list exactly.