Staff AI Engineer (Agents)
- Hiring from
- Australia
- Work type
- Hybrid
- Posted
626,690 remote jobs, straight from company career pages
100% free · New jobs every minute
Show job descriptionHide job description
Staff AI Engineer (Agents)
The Role
We're hiring Staff AI Engineers to build the agents at the centre of Lyrebird's clinical AI. You'll join Core Scribe, Automate or Integrations and own an outcome end to end: the note, the workflow after it, or the EMR connection underneath. Every output reaches a clinician, so it has to be right.
You've built agents for real, not demos. You know what breaks when an LLM meets production. Nobody here will hand you a spec: you'll find the problem, sit with the clinician who has it, decide whether it needs an agent or plain deterministic code, and defend that call. The way you build has changed as models have improved, and you treat that as a design input. You use AI to move fast, but everything you ship carries your fingerprint: no vibecoding, no slop, nothing a clinician pays for later.
About Us
Every day in Australia, 10 people die from preventable medical errors. Not from incurable diseases, but from misdiagnosis, missed information, and clinicians making decisions under impossible pressure. At the centre of that failure is a doctor spending 40% of their day on paperwork instead of patients. Lyrebird is fixing that.
We turn real clinical conversations into accurate notes, automatically, backed by ambient real time document creation, intelligent document management and clinical workflow automation, native patient data integrations across siloed health systems, and a patient and clinical communications layer. Thousands of clinicians across multiple disciplines use Lyrebird every day in environments where accuracy, safety, and time matter deeply. We're growing fast across Melbourne, the UK, and Asia, and there's a finite window right now to set the standard for clinical AI in Australia before anyone else does.
What you'll do
Own a product outcome end to end, whether that's activation and note quality, workflow conversion, or integration depth and reliability
Design, build and run production agents, from architecture and tool design through evals, rollout and the failures you only see in real clinics
Define what good looks like before you build, writing the specs and evals that make quality and hallucination measurable rather than anecdotal
Build reusable pieces so the next workflow or integration ships as configuration, not a rebuild
Partner with product as a peer on what to build, and make the smaller product calls yourself
Work directly with clinicians, clinical safety, EMR vendors and enterprise customers to find the problems worth solving
What you'll bring
Deep, hands-on experience building agentic systems, including at least one agent you took to production and owned through real usage, failures and iteration
A habit of writing evals as part of building, not after. You can explain how you defined good, what you measured, and what you changed as a result
Strong judgment on agent design: tool use, orchestration, context, guardrails and human-in-the-loop, and knowing when not to use an agent at all
Daily fluency with AI coding agents like Claude Code or Codex, holding what they produce to the same bar as your own code
Product judgment: you can say what you chose not to build, and why
Nice to have
FHIR and HL7 in practice, ideally having felt the pain of integrations built by hand
Experience in healthcare, SaMD or other safety-critical environments
Time spent working directly with end users, ideally clinicians, to shape what gets built
Familiarity with TypeScript and AWS
An agent that gets a clinical note wrong isn't a bug report. It's a decision a doctor makes on bad information. The systems you build will decide how much time clinicians get back and how far they can trust what's in front of them. That's why the bar is this high, and why the people who meet it care about it this much.
We're building a team that reflects the diversity of the people who benefit from our work. We want Lyrebird to be a place where everyone feels safe, supported, and able to thrive. If you're from an underrepresented background in tech, we strongly encourage you to apply, even if you don't meet every requirement.
