Flagship · illustrative composite · 01 / HMS · Jan 2022 — Jul 2026

Referral status API

Flagship case. An illustrative composite of HMS healthcare work: the operator screen that asked whether a referral had moved — Java, Spring Boot, SQL, React, AWS.

35% faster APIs · 40% shorter deploys

Illustrative preview of Referral status API — Illustrative preview

Problem: Healthcare workflows stalled on slow APIs and heavy SQL. A change took too long to reach production, so the team batched work and waited.
Constraints: Existing Spring Boot microservices, production traffic, and integrations that could not be paused for a rewrite.
Decision: Keep the service boundaries. Spend the work on API design, SQL, and a Jenkins + Docker pipeline so a release was a habit instead of an event.
Tradeoff: Calendar time went into design conversations that delayed some features. Those two résumé numbers are the ones I would still defend in a room.
How it was measured: 35% faster API response times and 40% shorter deployment time — the résumé numbers. The referral-status framing is an illustrative composite of that HMS work, not a production runbook.

Where the wait lived — illustrative

React operators
Spring Boot APIs
SQL — bottleneck: status reads

Same service boundaries. The wait was in the query, not a new service.

Healthcare software has a particular silence to it. Nobody thanks you when a referral moves. They only appear when the status screen takes too long for someone who is already tired.

At HMS that screen was referral-status: React for the operator, Spring Boot in the middle, SQL and AWS underneath. The expensive moment was not the late change. It was the early silence — when everyone felt the slower query and nobody said it.

I said it, then proved it in the plan and the pipeline. The service kept its name. The wait on the screen did not.

Java
Spring Boot
SQL
REST APIs
React.js
AWS
Docker
Jenkins