[stokedbits]

Updated the architecture for the deployment model for millions of hospital devices. It originally required service techs to take a thumb drive to a hospital with the update and manually using a series of installers and scripts to perform tedious steps that could take days to update one device… usually with errors because the steps were not followed correctly.

The update was to put all the build steps into source control, decouple all the logic, setup ci/cd to support the changes, and use a series of aws services (greengrass, kinesis, S3, lambda, cognito, systems manager, app sync, Api gateway, cloudfront, dynamo, etc) to facilitate an event driven architecture. Then threw an easy to easy to use gui (nextjs/react) on top of it for the customer support teams to use that was as fail proof as possible with a holistic view of system state and update progress.

Easily saved the company 10s of millions of dollars yearly and update times for the most critical systems went from days to hours. Failure rates went from a problem to a minimal occurrence which we had the observability in place to resolve in future scenarios by adding additional tests for any regression issues found.

Super fun project, learned a lot about AWS. Made a lot of customers happy and hospitals more safe.