Context
Photon built secure communication for on-call physicians: messaging, patient data, and medical imaging shared between clinicians, with a patient side to the same conversations. In clinical messaging, ambiguity costs a wrong call about a patient. It had to be fast and unambiguous on three platforms, under HIPAA, for two groups who want opposite things: clinicians want density and speed, patients want reassurance and plain language.
The company had engineers and a working backend. No designer, no system, no research, no process for getting design in before a feature was built. I was the first design hire, April 2023 to July 2024, and the only one.
Contribution
Practice before product. The first weeks went into one component library across iOS, Android and mobile web so the three engineering tracks stopped drifting apart, and into a research cadence: clinician interviews and patient sessions, the company’s first. The founders were used to seeing screens every week and got a system instead. That was the cost.
The product call that mattered: split clinician and patient at the pattern level, not the screen level. Same conversation, two renderings. The clinician sees the thread with patient identifiers, imaging thumbnails and timestamps at full density. The patient sees the same thread with identifiers stripped, type enlarged, and the next action up top. One data model, two component sets. Maintaining two sets is more work. The alternative was a compromised version of both.
I added A/B testing and worked directly with the iOS and Android engineers through implementation, so what shipped matched the spec at the component level.
Outcome
A live clinical messaging product on three platforms, running on the system and research practice I set up.