Design System
OVERVIEW
I led the creation of a scalable, multi-brand design system for Marriott’s Global Learning & Development (L&D) organization. The system supported 30 distinct brands, each with different visual identities, across web, tablet, and mobile platforms.
While this work predated today’s AI-driven experiences, the system was intentionally designed to handle high variability, rapid iteration, and cross-team adoption — the same foundational requirements of modern AI-ready design systems. The goal was not just consistency, but flexibility at scale.
The Problem
The L&D design team operated with extreme speed and limited resources, designing multiple products in parallel across dozens of brands. This created three systemic issues:
Fragmentation – Similar UI patterns were implemented differently across products.
Inflexibility – Brand-specific designs made reuse difficult and slowed iteration.
Poor design–engineering alignment – Developers often had to interpret intent, leading to rework.
In today’s terms, the system was not “future-ready”: it could not gracefully handle variation, growth, or change — the same challenges AI introduces to product experiences.
Design Process
As the system owner and librarian, I approached the work as a governance and adaptability problem, not a component inventory.
Key principles that directly map to AI-ready systems today:
1. Designing for variability
Components were built to flex across fonts, colors, and brand expressions without redesign.
Patterns prioritized structure over decoration, allowing content and styling to change independently.
This mirrors modern AI needs where content length, tone, and confidence vary dynamically.
2. Token-based thinking before tokens were mainstream
Styles were abstracted so brand changes could be applied system-wide.
This enabled rapid experimentation without fragmenting the system — a core requirement for AI experimentation today.
3. Cross-functional alignment as a system feature
I partnered closely with engineering to ensure components mapped cleanly to implementation.
The system reduced ambiguity for developers, which is especially critical when AI introduces non-deterministic behavior.
4. Adoption-first mindset
The system evolved in parallel with live products.
Designers used it daily, which surfaced gaps early and prevented “theoretical” patterns that break in reality.
Course Cards At All States
Table Components
Defining Spacing Between Elements
Solution & Success Measure
The system was built from scratch in under six months and became the default design standard for new L&D initiatives.
Outcomes:
Significantly reduced design and development turnaround time
Enabled consistent experiences across 30 brands and multiple platforms
Supported long-term evolution without needing redesigns from scratch
Became the foundation for Marriott’s first centralized learning platform