Full-Stack Software Developer - AI & Agentic Systems
Build the AI-powered platform our rehabilitation consultants use to manage every case, end to end.
Software Development & Applied AI
Mid-level — 3+ years shipping production software
Build the AI-powered platform our rehabilitation consultants use to manage every case, end to end.
Software Development & Applied AI
Mid-level — 3+ years shipping production software
Founded in 2019, AusRehab is an independent provider of workplace rehabilitation and injury management services. We work with employers, insurers, and allied health professionals to support early, safe, and sustainable return-to-work outcomes
We are growing across NSW, WA, and ACT and looking for people who want purposeful work, strong support, and the freedom to do their job well
Agentic engineering: designing agents and workflows that extract, reason over, and act on real case data. This is the centrepiece of the role. The hard part is not calling a model; it is building something reliable enough that a consultant trusts it with a file containing information that could affect a person’s livelihood.
Full-stack product development: Python backends, TypeScript and React front ends, mobile development with React Native, PostgreSQL, and the telemetry and reporting needed to tell us whether any of it is working.
Owning your infrastructure: AWS and GitHub Actions. There is no operations team to hand this to, which means real ownership and no ticket queue between you and production.
Talking to the people who use it: you will sit with rehabilitation consultants, administration staff, and accounts staff to work out what to build, then train them to use it. You will not be handed finished tickets. Requirements here emerge from conversations.
Production Python and production TypeScript, including React or Next.js. Both, not one.
Relational database experience. PostgreSQL is preferred, but experience with SQL Server, MySQL, or Oracle transfers well.
Real AI fluency, in two distinct senses: you have designed an agent or LLM workflow to solve an actual problem, and you use AI coding tools effectively. Both depend on the same thing: the judgment to review what a model produced and explain specifically why it is wrong. This is not vibe coding.
Solid Git discipline.
Communication that adapts: you may need to explain the same decision to a director, another developer, and a rehabilitation consultant in the same week, without dumbing it down for any of them.
Comfort with ambiguity, along with the honesty to flag early when something is harder than you expected.
No workers’ compensation knowledge is required. Nobody arrives with it. What we do look for is curiosity about a domain you have never worked in before: asking how the scheme works, what the data looks like, and why it is messy. The domain is learnable; disinterest is not.
If you want a career where your work has a real impact, and your expertise is valued, we would love to hear from you. Apply via our form below.