Forward Deployed Engineer

Summary Hatch engineering culture is driven by speed, ownership, and impact. We're a small, ambitious team building AI that transforms how service businesses grow, and we move fast. Engineers own problems end to end, ship to production early, and work across a modern stack (Elixir, Python, Go, React + TypeScript) on problems that directly shape the customer experience. At the end of the day, we're here to build something that matters, grow as engineers, and do it together.   The Forward Deployed Engineer is a first-class builder who ships production code into Hatch's core codebase, internal tooling, and the integration layer that connects Hatch's products to client systems. The primary focus is Hatch's Voice AI product: real-time voice agents that answer calls, qualify leads, and book appointments directly against clients' CRM and field-service systems.   "Forward deployed" describes where you work, not the caliber of what you build. You sit close to our most strategic clients, watch their real call traffic, and debug their failures yourself, working through agent traces, LLM tool-call payloads, service logs, and the client's own API until you can name the actual defect. Then you fix it where it belongs: in the client's configuration, in the integration layer, or in Hatch's core voice pipeline so every customer gets the fix.   Most of your week looks like engineering: reading traces, writing repros, instrumenting what nobody measured, shipping PRs, and validating them on production traffic. The rest is being the person both our clients and our Product team trust on the technical detail. You'll be deep in the build, held to production-grade standards, and given a genuine seat at the R&D table.   This opportunity is Hybrid working from our NYC Office (Soho)   We’d love to have you apply, even if you don’t feel you meet every single requirement in this posting. At Hatch, we’re looking for great people, not just those who simply check off all the boxes.   What you'll do   - Root-cause voice AI failures and ship fixes into the core product. Start from one bad call and work backward through agent spans, tool-call arguments, application logs, and warehouse queries until you can explain exactly what happened. Then push PRs into Hatch's voice pipeline and integration layer, gate risky changes behind feature flags, and confirm on live traffic that the fix actually fires. - Build and extend the integration layer. Direct API integrations, webhook endpoints, event-driven triggers, and connectors for CRMs and scheduling systems that have no Hatch integration yet. Reconcile what a client's business needs against what their API actually returns, including the cases where those two disagree. - Build the tooling and instrumentation that makes this repeatable. Local simulators, harnesses that replay production calls end-to-end, behavioral test suites that catch regressions before a rollout, plus the spans, metrics, and dashboards that…