Case study · Healthcare
HomeNurse: running a home-care operation on one platform
Built for Nachi Medical Solutions, a South African home healthcare provider. In production, serving the organisation's day-to-day operation.
The problem
A home healthcare provider runs an operation that is, by nature, spread out: nurses travel to patients' homes, coordinators assign visits, managers need to know what is happening today and finance needs to know what happened last month. Without a system built for this, the operation lives in phone calls, messages and spreadsheets — and patient information ends up in places it should never be.
Why it was hard
Healthcare data raises the bar on everything. Patient information has to be minimised, access-controlled and accountable under POPIA. The people using the system most — nurses — are on personal phones, often on poor connectivity, in someone else's home. The organisation has no internal IT team to run servers. And the platform had to be affordable to operate for a provider whose income does not scale like a tech company's.
The decisions that mattered
Five roles, not one admin flag
Managers, coordinators, nurses, finance and administrators each see and do different things. Access is modelled as explicit roles enforced on the API, not hidden in the front end — so a nurse's phone can never reach another nurse's patients, and an auditor can read who did what.
MFA on every account
Nurses sign in from personal phones in patients' homes. Every account requires time-based one-time codes on top of a password, and sessions are short-lived. New nurses register themselves and are only activated once a manager approves them.
Serverless, so there is nothing to patch
The provider has no IT team. The platform runs on managed AWS services — Lambda, DynamoDB, Cognito, S3 and CloudFront — so there are no servers to maintain, the running cost scales with use, and a quiet month costs almost nothing.
Two environments, one pipeline
Separate development and production environments, defined entirely as code with AWS CDK and deployed through GitHub Actions using short-lived OIDC credentials. Nothing is changed by hand in production, and every change can be traced to a commit.
Reporting built in, not bolted on
Statistical dashboards and PDF reports are generated from the same data the operation runs on, so management sees the real state of visits, nurses and patients rather than a spreadsheet assembled at month-end.
What we built
- Patient management with secure, role-scoped records
- Nurse onboarding with manager approval
- Nurse-to-patient assignment workflows
- Live monitoring of home visits as they happen
- Statistical dashboards for management
- PDF report generation
- Cognito authentication with TOTP multi-factor
- Infrastructure as code (CDK) with GitHub Actions CI/CD
What changed
The organisation runs its home-care operation on one digital platform: patients, nurses, assignments, live visits and reporting in one place, with every action attributable to a person and a role. There are no servers to maintain and the running cost tracks actual use. We continue to support the platform in production.
Building something in healthcare?
Tell us about the operation you need to run and the data you need to protect — or read how we build healthcare software.
Talk to an engineer

