All case studies

Mr. Goma Tires · Automotive · 2024 — present

Selling used tires like premium products

E-commerce for 7 Florida locations with tire-size search, a faceted catalog, 3D tread views, Stripe checkout and a sales dashboard with an AI chat that filters inventory in plain language.

down from 0.914, plus mobile LCP from 4.5 s to 3.2 s on the product page
CLS → 0
Type
Freelance
Stack
  • Next.js
  • React
  • TypeScript
  • Tailwind
  • Three.js
  • SQL Server
  • Stripe
  • Claude API
  • n8n
  • Vercel
Status
View live (opens in a new tab)

The business

Mr. Goma Tires sells used and new tires and offers auto services at 7 locations in Miami and Orlando. It doesn’t compete on catalog size or price, but on experience. The project’s mission: make buying a used tire online feel as clear as buying a new premium product.

What I built

I led and built the platform end to end, in English and for US shoppers.

  • Storefront: tire-size search from the homepage, a catalog with faceted filters and counts, brand and size pages, and a product page with a gallery, zoom and a 3D tread view showing depth in 1/32” and % of tread life left.
  • Checkout: cart and checkout with Stripe, WhatsApp orders as an alternative, and a step-by-step quote tool connected to n8n.
  • Sales dashboard: an inventory table with filters by location, plus an AI chat (Claude, with tool use). Salespeople type what they’re looking for in plain language, and the model applies the table filters.
  • SEO and analytics: a Google Merchant Center feed, structured data, guides with FAQs, pages for each location and service, and consent-based analytics.

Decisions that mattered

  • Spec-driven development with Claude Code: over 35 documented features, each with its own spec, plan, tasks and measured results.
  • Saying no with data. An SEO audit recommended consolidating “duplicate” URLs. The analysis showed those units had real price differences (a 22% median), so we kept them separate.
  • Performance on a budget: a JavaScript size limit enforced in CI and hundreds of automated tests.
  • Security in production: we detected and mitigated a bot attack of 150,000–200,000 fake visits a day.

The result

  • CLS from 0.914 to 0 on mobile and from 0.525 to 0 on desktop, on the product page.
  • Mobile LCP from 4.5 s to 3.2 s on the same page.
  • The gallery zoom went from taking 1.7 s to being ready before hover, with images cut from 283 KB to 90 KB.
  • 29% lower daily infrastructure cost after mitigating the automated traffic.
esc