Automated provider matching
IntelliWake, 2025
65%
of jobs now assigned without a dispatcher
- Problem
- Dispatchers hand-picked a driver or interpreter for every medical appointment. It was slow, inconsistent, and didn’t scale with volume.
- Result
- Automated assignment for 65% of jobs, saving dispatchers about 12 hours a week.
- Stack
- TypeScript
- Node.js
- PostgreSQL
- React
- Scheduled jobs
- My role
- Designed and built the matching engine as one of several engineers on a mature production platform (612 API endpoints, 234 database tables).
What I built
- A matching engine that ranks eligible providers by a weighted score and quality tier, and checks more than 15 eligibility rules: credentials, suspensions, capacity limits, coverage area, contract coverage and client opt-outs.
- A nightly job that re-scores providers and places them into quality tiers, with a manual override that records who approved it and when it expires.
- A simulation tool that lets the operations team test rule changes against real data before they go live.
- Admin settings for distribution weights, four operational reports, and unit tests for the matching rules.