During the implementation of the requested features, the Lohika team realized that the existing architecture and codebase were error-prone, which would slow down the development and overall quality of delivery. As soon as we achieved feature parity and successfully released the app for the iOS and the Android platforms, the Lohika team proposed a plan for a major refactoring of the application. The proposal included a rebuild of the backend system and a database schema redesign to eliminate bugs and speed up development based on existing and future usage scenarios.
The team successfully implemented the refactoring changes that included a database schema redesign, numerous bug fixes, a new architecture design and an update to the latest essential frameworks versions.
The new architecture is completely based on AWS managed services. The mobile application interacts with a serverless backend based on AWS Lambda functions in JavaScript, talking to AWS RDS/Postgres. Authentication is done with AWS Cognito. Communication with ML services owned by our client’s ML team is made through AWS SQS and Kinesis streams.
Our team built a Jenkins-based automated CI/CD pipeline from scratch. It includes code style checks, third-party library vulnerability scans, database migrations, provisioning, configuration management, and deployment of different environments. Lohika QA engineers performed API testing for endpoints, with additional checks for various non-trivial edge cases.
Execution of these API automated tests are now a mandatory part of the CI pipeline. This initiative eliminated a large number of bugs and was used as a safety net during further development, when the existing logic was frequently re-factored. As a result, our client now has a product with significantly improved test coverage that continues to evolve.
The key features implemented by the Lohika team are as follows:
- A new version of the tracker with continuous heart rate, glucose data, food, activities and other logs in one place on the timeline
- Visualized progress of consumed nutrients on the health dashboard
- History of food, activity, medication, water and weight logs
- User feed providing data insights on how consumed food or performed activity affects user’s glucose and heart rate
- Food search by user photo leveraging Clarifai solutions
- Premium paid version of the app
- And many others
The Lohika team was also responsible for integrations with wearable devices, such as Apple Watch, Fitbit, Mi Band, Abbott FreeStyle Libre, and Dexcom to track users’ HR and CGM.
Since the app is a healthcare product, it has to comply with HIPAA. Our team delivered several features to achieve HIPAA compliance, such as protecting access to the app with touch/face ID or with a PIN code.
The app only uses secure storage for saving local data on a device, and for enhancing the backend infrastructure. The latter included data de-identification, preventing all sensitive information from being logged, and using only HIPAA-compliant third-party services.