SAP leadership wanted to stop shipping scattered machine learning tools and start shipping one integrated platform, seamless end to end. I joined the Machine Learning Foundation in January 2019 as UX Lead for that consolidation, responsible for consistency across roughly ten engineering teams in seven cities (Berlin, Walldorf, Dublin, Tel Aviv, Singapore, Palo Alto, Bangalore), building on three different frameworks: SAPUI5, Angular and Aurelia. SAP's Fiori guidelines existed, but they were broad enough to interpret ten different ways, which is functionally the same as not having them.

Decisions and Actions

• Built a design system from scratch, referencing Fiori 3.0, with the reasoning behind each decision written down next to it, not just the rule. That written reasoning is what let designers and developers who had never met decide things the same way.

• Ran design critiques across the platform's teams, naming the inconsistent UI and interaction patterns, then experimented with reducing how many critiques were even needed, since a written single source of truth answers most of what a critique exists to catch.

• Got product owners to agree to validate features ahead of the development cycle, in an organisation where testing before building was not yet the norm and most engineers had never had a UX designer on their team at all.

• Recruited consultants closest to our Beta customers for early interviews, specifically because they could surface real pain points faster and cheaper than a full formal research cycle, then ran task-based feedback sessions with the data scientists and ML engineers actually building on the platform.