The work falls into four areas. Most engagements draw on more than one.
I design the clinical logic an AI runs: conversational workflows, decision points, escalation rules, and data capture.
I do the Clinical Safety Officer work under DCB0129 and DCB0160: identifying hazards, designing controls, and building the safety case.
I translate clinical need into something a team can build, and judge when a feature is safe enough to ship.
I test whether the system does what it should and fails safely when it does not, and produce the evidence to prove it.
Healthcare AI is becoming more autonomous. As these systems move up the levels of autonomy, the clinician sits further from each decision. The human stays in the loop, but oversight shifts from checking individual outputs to the design of the system itself. That turns safety from a review at the end into a design problem from the start.