Laika Replatforming

Context

Laika's storefront was a PHP monolith without i18n. Mexico and Chile were blocked by the architecture, not by a missing translation file.

My ownership

• Technical lead for architecture and delivery inside a 16-person cross-functional squad (engineering, QA, design, product).
• Next.js, React, TypeScript and GraphQL storefront: catalog, checkout, membership and loyalty.
• i18n design for CO, MX and CL, and the Chile launch.
• Handover: documentation and training so the platform stayed maintainable.

Constraints

• Could not freeze the business for a rewrite. Incremental cutover.
• Three locales, three commercial realities, one codebase.
• SEO mattered for a storefront. SSR was not optional.

Architecture

PHP monolith  -- strangler -->  Next.js storefront
                                 React + TypeScript
                                 GraphQL (Apollo)
                                 i18n: CO | MX | CL
                                 SSR for catalog / landing

Technical decisions

• Strangler path instead of a big-bang rewrite.
• GraphQL to stop over-fetching the catalog the way the PHP pages did.
• Locale from domain / config, not a client-only toggle.

Tradeoffs

The PHP monolith and the Next.js storefront ran in parallel during the cutover. Traffic moved by locale, not in one night.

Impact

• Chile and Mexico launched on the new platform.
• Launch analytics showed a 40% engagement lift versus the previous Chile storefront.
• SSR pages that search engines can read.

What I learned

• i18n was a catalog and routing problem. Without locale on the storefront, MX and CL could not launch.
• Technical lead covered architecture and delivery for the squad.
• The 40% figure is the Chile launch comparison against the previous storefront.