Does redefining what Quality looks like in the age of AI — on a brand-new AI product — make you tick?

At Re-Leased our vision is to shape the future of Commercial Real Estate and drive better outcomes for people, property and planet. We're now building am:pm — an agentic AI product for commercial real estate operators.

Property managers run their portfolios on a system of record like Re-Leased, but most of the actual work happens outside it — emails, phone calls, follow-ups, to-do lists scattered across tools, with no system behind them. am:pm is the system of action for that work: it captures the work to be done as jobs and tasks, grounded in the communication channels, memories, and documents that surround the portfolio — so the work itself finally has a home. You can read more at ampm.ai.

am:pm is in early access with customers right now. The squad is small, the stack is new, and we're moving fast.

Your Role

We're looking for a senior quality automation engineer to own the quality system for am:pm — based in Napier or remote from another city in New Zealand. The whole squad owns quality; you build the automation, quality controls, and practices that make that possible. This is an automation-led role: engineers write unit and integration tests, and you shape where automated coverage sits across the test pyramid and steward the end-to-end suite, with AI as a core part of how you work — generating tests, triaging failures, creating test data. We're also moving to continuous delivery, and a core part of this role is making that safe — the right level of automated quality controls so changes flow to production with confidence.

And because am:pm is an AI product, quality means testing agentic behaviour: non-deterministic outputs, tool use, statistical rather than binary correctness. Engineers write the evals — just as they write unit and integration tests — and you bring the quality lens: when evals are the right control, and what they need to catch.

This is a small-team, start-up environment inside an established SaaS business — less process, more agency, and a tight feedback loop with real customers.

Your Responsibilities

Drive quality automation across the test pyramid — set where coverage sits so tests run at the layer that's cheapest and most reliable — and steward the end-to-end suite: curate what earns a place in it, and set the standards engineers follow when contributing tests of their own

Support our move to continuous delivery — design the right level of automated quality controls (pipeline gates, smoke checks, release signals) so we can release frequently without slowing the flow of change

Use AI tooling as a first-class part of your workflow — test generation, bug triage, test data creation — and keep raising the bar on how the team uses it

Curate AI-generated tests and findings — decide what to keep, reject, or refine, applying your own testing judgment rather than accepting output wholesale

Shape how we test AI-powered features — advise on when evals are the right quality control and what they should cover, and contribute adversarial and edge-case scenarios; engineers write and own the evals themselves

Partner with engineers on acceptance criteria and coach them on designing test scenarios from a user's perspective — positive paths, negative paths, edge cases

Run exploratory and risk-based testing to find what automation misses, and feed production bugs back into automation at the layer that would have caught them

Analyse quality workflows — triage, release, escape defects — and drive structural improvements, reporting on quality signals so the squad always knows where quality stands

Your Experience

5+ years of QA experience in a SaaS product environment, with automation at the centre of how you work

Track record of designing test automation frameworks from the ground up — architecture, tooling choices, conventions — and building the JavaScript/TypeScript end-to-end suites (e.g. Playwright, Cypress) that run on them

Comfortable working in a production codebase — Git, branches, PRs, code review — and able to read code well enough to locate and understand the surface under test