Engineered a multi-tenant, white-label SaaS platform end to end: Express/MongoDB Atlas backend, a Next.js 16 admin/analytics app, and a member-facing provider-search UI — workspace-scoped so each health plan's data stays isolated
HealthcareWe built and run the platform that keeps Medicare Advantage provider directories accurate — and keeps plans out of CMS audit trouble.
Healthcare / Health Insurance (US Medicare Advantage payers)
insurelytix.app
Directory accuracy
- Provider record 0148Match
- Provider record 0149Match
- Provider record 0150Stale
- Provider record 0151Match
- Provider record 0152Phantom
Accuracy
96.4%
CMS adequacy
Pass
The challenge
US Medicare Advantage health plans must prove to CMS that their provider directories are accurate and their networks are adequate, but most small and mid-size plans run this on spreadsheets and $100K+ consultants — leaving ghost-network listings, audit exposure, and quarterly adequacy fire drills. InsureLytix needed a payer-grade, multi-tenant product built and run in production, not another slide deck.
Healthcare
Healthcare / Health Insurance (US Medicare Advantage payers)
What we built
The system, in parts.
Built the directory-accuracy program: NPI registry cross-checks, Match/Stale/Phantom row classification, decay alerts and remediation queues to hold large directories close to ground truth (Liberty MA ~30K providers, Clear Spring Health ~59.8K providers across MA + D-SNP)
Built the CMS network-adequacy and gap-analysis engine — drive-time/time-and-distance compliance against CMS standards using provider GeoJSON + Google Maps, plus county-level pass/fail scoring and a what-if disruption simulator
Turned messy payer documents into structured data: an async CSV ingestion pipeline (validate → insert → score) with row-level validation, AWS Step Functions audit jobs, and OpenAI-assisted schema validation that maps unstructured uploads onto a clean provider/facility/pharmacy/formulary model
Shipped FHIR Plan-Net APIs (X12 837/834 bridges where needed) so plans meet the CMS Interoperability Rule without building it in-house
Added OpenAI-generated network-health narratives, a WCAG 2.1 AA / Section 508-compliant member search, and HIPAA-aligned infrastructure with BAA and immutable 7-year audit logs
Outcomes
What changed for them.
Live in production for real Medicare Advantage plans since 2023 — provider directory and FHIR Plan-Net running on each plan’s own domain
No directory-accuracy audit findings reported for the flagship plan, turning a recurring compliance liability into a managed program
Replaces six-figure point tools (e.g. Quest Analytics-tier adequacy software) with one operational source of truth across member, broker, and ops surfaces
Quarterly network-adequacy reporting shifts from spreadsheet fire drill to an always-on, CMS-aligned scorecard
Customers chose to sign direct contracts to keep the product when their channel partner wound down — a clean signal of product-market fit
Offered two ways to consume: most plans hire the team to operate the directory, some license the platform to run it themselves
How it’s built
The stack.
Next.js 16 (App Router)React / TypeScriptExpress + MongooseMongoDB Atlas (multi-tenant)AWS S3 / SES / Lambda / Step FunctionsFHIR Plan-Net (X12 837/834)Google Maps Distance MatrixOpenAI (schema validation + network-health insights)StripeRedux ToolkitTailwind
More work
See allTwo ways to start
Want a system like this?
Tell us the workflow you want to run itself. We will scope a focused first project — designed, built, and operated, with humans in control.