Klaviyo mockup

Klaviyo

Klaviyo is an AI-first B2C marketing/CRM platform that helps consumer brands turn their customer data into personalized email, SMS, and push campaigns. It supports fast-growing DTC customers like Glossier and Liquid Death to household names like Mattel.

Year

2026

Role

Product Design Co-op

Skills

Design Systems, Design Engineering, AI

Team

Core Infrastructure


The Problem

Ascent is Klaviyo's enterprise design system. My contributions to the system include a carousel incentivized by growing screen density, and new content organization patterns built to accommodate enterprise customer needs.

The Solution

I worked as a product designer with codebase access, which meant I owned the work the whole way through spec, code, review, and into production.

Key Contributions

Reduced code redundancy and maintenance. Cut down on six custom carousel implementations and closed the accessibility gaps that all the custom one-off versions had.

Tackled legacy UX debt for scaling customers. Enterprise customers needed a sturdier way to manage their account assets, so I reworked the patterns for a folder-based design pattern that introduced hierarchical categorization.

Owned to production. 10+ Figma specs, taken all the way to a merged beta-library component with full test coverage.

Sorry! Some of this work is proprietary. Please reach out for further details

Outcomes

Total contributions touched more than 56% of Klaviyo's app surfaces.

Takeaways

Owning a component from Figma spec through code review taught me to design with the build in mind. Knowing what's expensive to implement helped tailor my specs to production patterns.

Closing accessibility gaps at the component level fixes them across every surface at once.

Retrospective

Migrating six teams off their custom versions outlasted my co-op scope. I built the component but I didn't get to finish moving everyone onto it.

Deep over broad categorization fits enterprise needs, but depth introduces more places to look, more clicks to navigate, and more friction along the way. It is a trade-off that needs additional validation and testing.