← All worknational HIE / interoperability

Designing and proving Ethiopia's national health data highway

Led the engineering behind a proof of concept for Ethiopia's National Digital Health Interoperability Highway, then turned it into a reusable platform other country programmes can build on.

Client
Ethiopia Ministry of Health
Role
Lead engineer
When
Apr 2026 – Aug 2026

The problem

Ethiopia’s Ministry of Health needed to prove out a national health information exchange (the NDHx) that could connect community health data from eCHIS to DHIS2, a shared health record, a master patient index and national identity. The work came through JSI and the Global Fund’s DHIA programme. The design also had to be reusable: other country programmes would want the same foundations without inheriting Ethiopia-specific decisions.

What I did

  • Built the NDHx proof of concept as lead engineer. It runs three interchangeable API gateways (OpenHIM, Kong and APISIX) on one orchestration bridge, with eCHIS-to-DHIS2 ingest on both the aggregate and FHIR tracker paths. Behind that sit a HAPI FHIR shared health record, the JeMPI master patient index, an OCL terminology cache, and Keycloak with eSignet national identity.
  • Implemented an OpenFN Lightning workflow path at parity with the custom mediator bridge, including a cron-triggered eCHIS pull and an OpenFN Kafka bridge. Together they made up a four-channel demonstration across two workflow engines, with both synchronous and Kafka-backed asynchronous delivery.
  • Extracted the proof of concept into an origin-neutral integration platform. I designed a platform/domain split so downstream country projects can take upstream fixes without conflicts, with routing defined in a gateway-agnostic YAML manifest that’s translated into each gateway’s native model.
  • Wrote MeshScope, a live service-mesh viewer that builds its topology from container labels and distributed-trace service graphs, so new services appear without any configuration.
  • Set the technical direction for the Kubernetes and GitOps phase (Helm, ArgoCD, Ansible and k3s bootstrapping on-prem), which three engineers then delivered on that foundation.

How I showed up

The brief was a proof of concept. I pushed it further so it would outlast the engagement. Swapping between three gateways and two workflow engines let the Ministry compare options against real traffic instead of slideware. MeshScope started as a tool to make the system visible to everyone in the room, not just its author. The team made it the platform default and later ported it to Kubernetes on their own. I set the direction for the next phase and handed it to three engineers to deliver, rather than holding onto it.

Outcome

The Highway was deployed into the Ministry’s sandbox environment. The extracted platform is now the base that other country programmes build on.