HWI Care — Full-Stack Provider Directory Platform

A national health-and-wellness licensing company needed a public platform to showcase a large, growing network of providers — findable by name, location, and specialty — and a way for non-technical staff to publish and maintain richly-branded content without a developer in the loop. Off-the-shelf directory tools crawl or collapse well before a network this size.

The directory has to stay fast at any scale — searchable by name, by city / state / ZIP-and-radius, and by specialty, rendered on an interactive clustered map — while the content behind it is fully staff-editable with drafts, publishing, and rollback. That means genuine geospatial search, a map that can never ship a multi-megabyte payload, a content system that never leaks a half-finished draft to the public, and a hardened, role-based back office.

A custom platform built from the data model up. Directory search returns in under a fifth of a second — load-tested past 100,000 records — over a scale-aware interactive map that ships individual pins for small result sets and server-side aggregates at scale. Staff manage every page through a custom page builder: draft previews, one-click publish, immutable published snapshots so the public never sees an in-progress edit, and full version history to roll back mistakes. A secure, role-based admin gives different staff different permissions with every change logged for accountability. SEO (rich metadata, structured data, sitemaps) and WCAG 2.1 AA accessibility are engineered in, not bolted on — and the whole thing is monitored, error-tracked, and kept running in production.

• Directory search under 200ms, load-tested past 100,000 records
• Custom page-builder CMS with draft previews, publish, and version-history rollback
• Role-based staff admin with every change logged for accountability
• SEO and WCAG 2.1 AA accessibility engineered in from the start

A live, in-production platform at hwicare.com that holds sub-200ms p95 on directory search — architecture proven against a 100,000-record load test, not against live volume. The public roster is currently curated to a small set by the client’s rollout choice; the point is that the system was engineered to stay fast long before it needs to. Staff run their own content day to day, and the platform is maintained in production with monitoring and structured logging catching problems early.