Yelp // 2020

Co-Creating an AI-Driven UI Framework

Designing a scalable MDUI system that allows machine learning frameworks to safely assemble the perfect card layout on the fly.

Yelp needed to dynamically customize search results UI based on personalization models. I co-created an AI-driven UI framework to handle hundreds of potential combinations safely.

-30%

Design-to-Dev Iteration Time

AI-First

ML-assembled dynamic layouts

Scalable

MDUI framework co-creation

Role

Lead Product Designer

Responsibilities

AI UI framework architecture
Scalable component constraints
Engineering collaboration
Interactive prototyping

The Challenge

Dynamic Personalization at Scale

Since the UI would be dynamic, we needed a system that functioned like a modern MCP-enabled design system. Each element was described with strict constraints, allowing the ML framework to assemble layouts without human intervention for every variant.

The Smart Framework

Merging Design Systems with Context

I designed the system to break down UI elements into functional components that the machine learning model could "understand" and place based on user and search-specific context.

“This system reduced design-to-dev iteration time by 30% through specialized search micro-components.”