The engineering challenge was structure. Film credits and professional identity are high-trust data - once a professional's record exists, it needs to be accurate, portable, and built to grow. Every architectural decision reflected that.

We designed a contract-first API using oRPC, establishing a strict boundary between frontend and backend that keeps both sides in sync as the platform scales. Authentication was handled through Better Auth, supporting multiple login strategies without coupling auth logic to business logic.

The data model was built for read performance and long-term integrity. Static reference data - departments, genres, industries - is denormalized at write time, storing labels alongside IDs so the read path stays clean without joins on volatile lookup tables.

On the frontend, we applied compound component patterns and vertical feature slicing - keeping UI composable, state dependency-injected, and business logic isolated per feature. Server components handle public reads directly, skipping unnecessary API hops while React cache() deduplicates database queries across the component tree.

A Mautic CRM integration runs alongside onboarding as a non-fatal side effect - capturing leads and pipeline stages without ever risking the core user flow.