How to Build an AI Medical Device Team Around the Product’s Intended Use
“AI” is a broad label. It does not tell a hiring manager what a person will build, what decisions the product will support, or what evidence the company will need. A useful hiring plan starts with the product function and the problem it is meant to address.
Once the function is written down, the company can decide which capabilities it needs and who will own each one.
Define the product function before opening a role
An AI-enabled device might identify patterns in an image, estimate risk from patient data, or produce an output that supports a clinical decision. Each function creates different technical, clinical, and regulatory work. The phrase “we need AI talent” leaves those differences unresolved.
The FDA’s draft guidance on AI-enabled device software functions recommends that manufacturers explain how AI contributes to the device’s intended use, what inputs and outputs it uses, and how the software fits into the intended user’s workflow. For hiring, that description helps distinguish a model-development role from work in regulated software, clinical evidence, product security, or systems integration.
Map responsibilities to capabilities
AI-enabled medical device work draws on several areas. Model developers work with data and performance. Software engineers build around requirements, architecture, verification, and traceability. Clinical and regulatory specialists establish what evidence supports the intended use. Quality professionals connect development to controlled processes, and product-security specialists address security risks.
Those responsibilities do not always require separate hires. The hiring plan still needs a clear owner for each one. Otherwise, important work falls through the gaps, and in AI-enabled devices those gaps tend to sit between model development and everything regulated around it.
Example: an imaging startup
Consider a startup building software that flags possible findings in radiology images for a clinician to review. Its first instinct is to hire an “AI engineer.” Before opening that role, it needs to decide what the person will own: the model, the data pipeline, performance testing across patient groups and clinical sites, or integration into regulated software. Someone also has to own clinical evidence, quality-system documentation, and product security.
If the model already exists and the gap is getting it into a controlled, traceable software product, the first hire is a medical device software engineer. If the model is still being built, a model developer comes first, and the software role follows once the architecture settles.
Ask candidates for evidence of the work
Titles and standards on a resume provide context, but employers still need to understand what a candidate personally handled. A candidate who says they “validated an AI model” should be able to explain what data they used, what the evaluation established, and how the results related to the intended use.
Interviewers can ask candidates to describe a product function they worked on, the decisions it supported, and the evidence used to assess its performance. They can also ask how the candidate handled a technical concern that arose during development.
How JBL Resources can help
JBL Resources recruits the regulated side of AI-enabled device teams, including medical device software, quality, regulatory affairs, and verification and validation. Every search starts with pinning down what the hire will own, and candidates are screened on the work they’ve actually done in that area.
If your team is defining a role for an AI-enabled medical device, download our guide, Hiring at the Intersection: A Practical Talent Guide to AI-Enabled Medical Device Software, or talk with our team about the search.
Comments