← All work

Building a shared design system across Figma, code, and AI

One naming system and one token structure for design, engineering, and AI-generated experiences.

Design SystemTokensNaming
View Figma exploration ↗
Problem

Features were being designed and built with different colors, type, spacing, and naming, so every new screen added more inconsistency.

What I did

I led the system from the base up: a naming convention agreed with engineering, a token structure introduced in steps, starting with spacing and then color, and a written record of where the platform sets the limits.

Result

Design and engineering share one naming system and one token structure, and the exceptions the platform requires are documented.

Role
Product Designer
Led the naming, tokens, and platform rules
Team
Design, Engineering
Platform
Figma, Shopify

Context

Mood did not have a shared design foundation, so features were being designed and built with different colors, typography, spacing, and naming conventions.

I started with the layer underneath the components: the values and names every component depends on.

The goal was a shared language that design and engineering could agree on, so every new component and pattern would build on the same base.

Before components, we needed a shared language.

The biggest problems were not missing components. They were inconsistencies underneath them.

01
Different values for the same visual purpose
02
Naming decisions made independently from engineering
03
Platform limitations discovered too late
04
Large changes difficult to validate across real screens

Three layers

01
Foundations
Color, type, spacing, opacity, and a naming convention agreed with engineering.
02
Token migration
Tokens enter the product in steps, starting with spacing and then color, so each change can be checked on real screens.
03
Platform constraints
Document where the product cannot reproduce the design exactly, and agree on the exception with engineering.

A design system still has to respect the platform.

Shopify Checkout

Shopify Checkout could not reproduce every design token exactly, including the original gradient and border radius.

Instead of treating this as a design failure, we documented the limitation and agreed on the closest supported implementation with engineering.

Outcome

Design and engineering now work from one naming system and a token structure with a clear order: spacing first, then color.

Spacing tokens went into the product first, which gave the team a real-screen check before each new set.

Where the platform sets a limit, as with Shopify Checkout, the exception is written down and agreed, so nobody has to rediscover it screen by screen.

What I learned

A design system is not only a collection of reusable values.

It is also an agreement between design, engineering, and the platform the product is built on.

The more clearly those constraints are documented, the easier it becomes for both people and AI systems to use the system consistently.

Hiring for growth or e-commerce design?

I’m open to product design roles in DTC, e-commerce, and SaaS. Email is the fastest way to reach me.

Next: Redesigning Quick Add to build confidence before checkout → ← All work