The autos suite started as a take-home exercise during the Cox Automotive interview process. The prompt: use a provided API to build a UI listing dealers and their available vehicles.
The provided server didn't support CORS, which made browser-based development impossible. Rather than stop there, the server was re-engineered from scratch and deployed independently to unblock the work. The UI was completed — with automotive brand logo icons added for each manufacturer — and submitted.
Before finishing, a GraphQL layer was built on top of the data and used as the foundation for a second, more sophisticated version of the exercise. Both versions were submitted to the recruiter. The extra work — circumventing a broken dependency, adding visual polish, and delivering a GraphQL API nobody asked for — was what solidified the offer from Cox.
Over the early years at Cox Automotive, the exercise was refactored and evolved into the reference app that now gets implemented for every new framework. The GraphQL backend runs on Vercel to this day.
7 implementations of a car-browsing app backed by serverless GraphQL APIs, with data based on racing games:
| App | Framework | Store | Visibility |
|---|---|---|---|
| autos-lit | Lit | nanostores | private |
| autos-p4t | Preact | Signals | private |
| autos-r3t | React 19 | Zustand+persist | private |
| autos-sld | Solid | built-in | public |
| autos-s4e | Svelte 5 | Runes | public |
| autos-vue | Vue 3 | Pinia | private |
| wc-autos | Stencil | @stencil/store | private |
Each: 10 components, action/state pattern, GraphQL backend, Tailwind CSS, SVG icons via dynamic web component loader.
Two serverless GraphQL APIs back the suite, both deployed on Vercel and built on car data from racing video games:
Using video game car data was a deliberate choice: rich, structured, publicly known datasets with manufacturer, model, year, and class data — all the fields needed to build a realistic dealer/vehicle browsing UI without any licensing concerns.