I was asked to redesign the compose box, one component. But I
found it behaved differently everywhere it showed up: InMail had
its own version, Pages had its own, Groups had its own. The real
problem wasn't the compose box, it was that messaging had never
been designed as a system. I audited 20+ compose variations across
surfaces, then convinced Premium, Pages, and Organic Messaging,
each with their own roadmap, to converge on one shared model.
I stress-tested it against two hero flows, messaging a
non-connection through InMail and messaging a connection for free,
starting with the compose box as the anchor and working outward
until the entire compose experience, message bubbles, media
attachments, reactions, statuses, banners, was rebuilt as one
coherent system. When LinkedIn began Nile, a multi-year
re-platforming of its core technical infrastructure, this model
became the blueprint for how the new architecture handled
messaging.