About The RoleSamsung Health helps hundreds of millions of people understand their health, through the devices already in their hands. We're looking for a
Staff Product Manager to lead the product definition behind a new AI coaching capability; the end to end experience, coaching logic, the behaviour model, and support the engine build and the practice for evaluating whether the coaching is genuinely good.
The problem takes real judgment: coaching that arrives at the few moments that actually matter, and stays quiet the rest of the time. You'll own that judgment layer.
This is a senior individual contributor role. You won't have direct reports, though other team members will look to you for day to day direction. You'll report to the Head of Product and work across teams in Europe, Asia and North America.
What You'll Do- Own the coaching experience. Start user first, and ensure the behaviour model and the decision logic serve what it feels like to be coached and determine what a coach says and, more often, when it stays quiet.
- Make decisions leave the room. Several organisations across several time zones build against what you define. Your job is to produce things others can act on without you present. Not documentation; thinking made buildable.
- Own the seams. Multiple interfaces across organisational boundaries. Contracts, schemas, who writes what, what happens when a signal doesn't arrive.
- Hold the evidence for coaching quality. Accuracy is the easy half. Helpfulness, personalisation, and whether a coaching moment feels like support are the hard half. You'd apply our thinking, extend it where it's thin, and own the evidence that the coaching is actually good.
- Decide what ships and what waits. Including what a first release is allowed to be wrong about, and what it isn't.
What You BringYou've taken something from undefined to launched. A product that reached real users while accountable for the outcome. You've invented a definition from an ambiguous starting point and can demonstrate the artefact you created to do it.
Product craft- You go into the material. You write, sketch and prototype your way to a definition rather than arriving with one. The brief describes what you found; it isn't the first thing you produce.
- You feel where an experience fails before the data says so - and you can name why it fails as a principle, not a preference.
- You author coherence. You can take a pile of features, screens, models, and constraints and turn it into one thing that feels intentional beyond a sum of parts.
- You treat AI product failures as context problems before model problems. What information reaches the model, when, in what form, and the minimum that produces an acceptable answer rather than the ideal one.
- You've killed something that worked. A feature that functioned, tested fine, and still wasn't earning its place. Knowing what to cut is part of product judgment.
How the work travels- Your writing does work. Specs, one-pagers, decision notes; these are how you think, not what you produce after thinking. Your artefacts survive being re-told into other organisations without you in the room.
- You start before the box is drawn. You don't wait for requirements, but you don't build blindly either. You'll decide at 60%, name the risk, and correct as needed.
- You anticipate what another organisation will need. Part of this job is being read correctly by people who may not sit in your meetings, in a different location, in a different working language.
- You've built ambitiously inside real constraint. A platform owner, a regulator, a parent company, an enterprise partner. You treat constraint as material to design with rather than a reason things don't work.
- You've worked more than one way, and you know which way fits. Discovery, delivery, evaluation, decision-making, practised in different company shapes, so you can say why a method suits this situation.
