Rebuilt the missing information architecture layer instead of continuing to patch the surface layer, then centralized the team around the corrected structure.

I proposed a bounded, reversible test to a business unit stakeholder rather than debating the question indefinitely: pause his agency's spend for a month and watch for negative impact. He ran it, found none, and we killed the spend for his business unit entirely — still no impact. That converted an open question into evidence.

Led a data visualization initiative applying a vector-based model to normalize behavioral data across the full portfolio, producing a shared visual language for evaluating design performance.

Drove department-wide adoption of a design systems model that went beyond UI components, converting every team's output into reusable component and pattern libraries across both horizontal and global capabilities.

Designed a single visual model that consolidated hundreds of use cases and multiple concurrent projects into one standardized deliverable.

Built a small, cross-trained creative team where each member operates across a wide surface of product and marketing work rather than narrow specialization.

Converted underused internal documentation into a visual taxonomy and shared vocabulary, deployed across every user touchpoint and reused internally to align product, engineering, and design.

Implemented a simple 5-step design-thinking process applied consistently across every operating cadence — daily, weekly, phase, and project level.

Applied a management-cybernetics approach, systematically running a design-thinking workshop across as much of the organization as possible.

Built cross-functional relationships deliberately and positioned the creative function as the rallying point for core initiatives, without any change in formal authority or mandate.

Deployed simple and pragmatic bits of AI capability across three distinct functions — research, management, and design — rather than as a point solution in one. Built the practice into daily operating cadence instead of treating it as a side experiment.

Built front-end coding fluency directly into the design function (vibecoding), closing the Skeleton-to-Surface gap the Elements of UX framework identifies as a common source of rework. Design output is now technically pre-validated before it reaches engineering.

I treated the practice as a system to design, not a website to build. First, I learned the market's words: I researched how buyers and recruiters search, and split their words into two lists that never mix in one asset. Second, I put my whole career in one spreadsheet — 18 engagements, 30 claims, 13 cases — and generated every case-study page from it, so a number exists in one place only. Third, I decided who does what: an allocation matrix, adapted from NASA Human Systems Integration, assigns each task to me, to the AI or to a script. If a script can check a fact, the script checks it. Fourth, every AI session reads the same rulebook first, and diagrams settle what words leave open. Last, I log every decision (43 so far) and audit drift every month.