I have kept the awkward parts in: the constraints, the wrong assumptions, and the decisions that changed direction.
The interface was only part of the problem.
A single case was spread across a desktop app, calls, spreadsheets, and delayed reports. I followed the work across three teams, then designed around continuity: what happened, who owns it, and what comes next.
Can someone pick up this case and know what to do next?
A worker should not have to stop and study the clock.
This shared Android terminal had to work on a wall, in bright spaces, with gloves, in English and Spanish, and across hardware that did not support the same inputs.
Did the punch work?
The product had to explain analytics while people were using it.
Audimate was a freelance project for people who knew spreadsheets but not SQL. I kept the data, chart controls, errors, and live result close enough for the workflow to stay understandable.
Can someone get from a file to a useful chart without learning the tool first?
The hardest decision was where the product should stop.
Health context had to sit beside the food choice without pretending to diagnose or replace medical advice. I rebuilt this student concept inside foodpanda's familiar ordering flow and documented its limits clearly.
How much guidance helps before it starts making the decision for someone?
TimeClock+ had years of interface decisions but no shared source of truth. I built the design system to help designers, engineers, and product teams work from the same patterns, then connected those patterns to AI-assisted prototyping.