MACH Headless Storefront
- Client
- Princess Auto · via EPAM
- Role
- Lead Software Engineer / Frontend Architect
- Timeline
- EPAM tenure · multi-year program
- Team
- Team · 14–18 engineers
Objective
Stand up a greenfield headless commerce storefront on MACH principles — Next.js 15, shared monorepo, and integration rails durable enough for a multi-squad team shipping to a global catalog footprint.
Decision log
Choices under review — why they stuck, and what they cost.
Turborepo + shared UI package over per-squad apps
Duplication was already taxing velocity; a single design system and build graph was the only way 14–18 engineers could move together.
Upfront platform investment delayed some feature starts — paid back in duplication cuts and review speed.
BigCommerce + Constructor.io with Azure Service Bus pipelines
Commerce and search needed event-driven boundaries, not a monolith of sync calls in the storefront.
Operational complexity shifted to messaging contracts — worth it for isolation and scale.
Storybook as the contract surface (50+ components)
Visual + a11y review needed a shared source of truth before PR merge, not after production screenshots.
Component discipline slowed cowboy UI — raised the floor for every squad.
What we built
- Greenfield Next.js 15 storefront on MACH boundaries
- Shared Turborepo monorepo and UI design system
- Commerce, search, and messaging integration rails
Panel observations · people
Set monorepo and design-system standards the wider program adopted
Mentored seniors into ownership of integrations and CI quality gates
Ran technical reviews that optimized for teachable decisions, not gatekeeping