The parts of the job that do not fit inside a single case study: how I lead, how I keep discovery honest, and the tools I reach for.
A continuous cycle, not a one-time research project. It is the rhythm I build into how a team works.
Talk to users on a cadence, not just at kickoff, so decisions start from evidence.
Put the smallest real thing in front of people and measure how it performs.
Turn what we hear and see into a sharp, shared point of view.
Refine the design against real feedback and ship the next version fast.
Tune the live experience toward the metric that actually moves.
I sit on an agentic workflow task force, helping bring AI workflows to the rest of the organization. I present my learnings to the design team and the broader product organization, and build reusable AI skills that give designers repeatable leverage. I work daily in Claude and ChatGPT, and give active feedback on the internal tools we are building.
I lead through standards and coaching. I build the operating model between design, product, and marketing, set delivery standards so quality does not depend on who picks up the work, and coach each designer where they are. I measure success by the team's output, not my own.
As Chair of Byte's Design System Quorum, I led the cross-functional design and engineering group that turned a fragmented design language into one org-wide system. I ran the audit, set the content models and naming, and drove adoption, so consistency became a shared standard instead of a personal effort. At Noom I brought the same instinct to the Nova design system, contributing token architecture. This is the leverage I care about: build the thing that makes everyone else faster.
I launched continuous discovery for our growth team at Byte and ran weekly user interviews to gather insights that impacted product strategy.
A slice of the interviews behind the practice. Each conversation was coded moment by moment, so the themes came from what people actually said, not from what we assumed.
The synthesis. Every interview moment coded into one shared tag taxonomy, then read as a distribution, so the team could prioritize on evidence: cost, candidacy, journey friction, and sentiment.