Machine Learning Engineer, Decision Tools

The thing that lets us triage 85% of patients off the orthopaedic wait-list is AI-supported decision support: models that help an Allied Health Provider (AHP) decide who needs a specialist and who does not. As a Machine Learning Engineer on Decision Tools, you build and maintain those models against audited clinical outcomes.

This is applied work on a problem where the cost of being wrong is real. You will partner closely with clinicians, ground every model in validated instruments, and keep humans firmly in the loop on every decision the system supports.

What you'll do

• Build, evaluate, and ship models that support triage and care-planning decisions, always as decision support a clinician can override.
• Translate validated instruments such as STarT-Back and the Oswestry Disability Index (ODI) into features the models can use responsibly.
• Stand up evaluation pipelines that measure model performance against published outcomes, including the 27-week to 4-week wait reduction.
• Work with clinical leads to define what good looks like and where a model must defer to a human.
• Keep the full pipeline auditable and within 100% Canadian data residency, from training data to inference.

What we're looking for

• 4+ years of applied machine learning in production, with strong Python and a solid grasp of evaluation and model risk.
• Experience deploying models that inform real decisions, ideally where interpretability and safety mattered.
• Comfort working with messy, regulated data and a habit of pairing every claim with the metric behind it.
• An evidence-led mindset; you read the literature, including triage studies such as Mutsekwa 2019 and Liddy 2020, before you model.
• Bonus: prior healthcare, clinical, or other high-stakes domain experience.

About CareChain

Most patients on a specialist wait-list do not need a specialist. We route each one through an AI-supported assessment led by an Allied Health Provider, so specialists see only the patients who need them and everyone else gets care faster. Our outcomes are audited and published: a median wait cut from 27 weeks to 4 weeks, with 85% of patients safely triaged off the list and 92% patient satisfaction.

We operate across 3 provincial health systems, 40+ clinics, and 2 academic centres, with 100% Canadian data residency. We are remote-friendly across Canada with hubs in Toronto and Montreal, and we offer health benefits from day one, equity for every full-time employee, and a vacation minimum we enforce.