Technology leadership · Learning and wellbeing
Rebuilding the foundations of an enterprise platform
A learning and wellbeing platform needed its applications, data and deployment foundations brought into a coherent system. An architectural review led to a rebuild of both principal applications and the iOS and Android apps, and a migration of the cloud estate.
- Role
- Led the redesign and carried out a substantial part of the implementation.
- Delivery status
- Both rebuilt applications and the iOS and Android apps entered production.
- Attribution
- Delivered in an earlier technology leadership role, before TechGuidr was founded. It was not a TechGuidr engagement.
The situation
The business was developing employee diagnostics, organisational analytics and AI-assisted briefings for leadership teams. The supporting estate had grown across different services, clouds and technology generations.
An architecture review found that some executive dashboards were disconnected from live production data and that calculation pipelines did not consistently implement the underlying methodology. Those were material dependencies for the product the business wanted to operate.
The decisions that shaped the rebuild
The applications needed a common data foundation. The rebuild connected transactional systems to a Databricks lakehouse, with calculation pipelines and a structured analytical model behind the executive views.
Removing an unnecessary middleware layer shortened the dashboard query path. The wider rebuild addressed the application, data model and cloud estate around it.
The AI Coach was grounded in curated, aggregated analytics. That decision set a boundary around the information supplied to it and made the data and calculations prerequisites for useful answers.
Before
- App generation 1
- App generation 2
- Detached services
- Second cloud
- Dashboards on disconnected data
- Scoring pipelines diverging from the methodology
After
Moving the cloud estate
The work included an Azure tenant-to-tenant migration covering identity configuration, subscriptions, app registrations, databases, storage, DNS and deployment pipelines. The migration consolidated the estate in the region required by the business.
A planned cutover maintained continuity during the move. Identity, access and deployment controls were part of the implementation alongside the applications and data platform.
Engineering detail
The revised architecture connected transactional applications, analytical calculations, executive reporting and the AI Coach. The diagram summarises the relationship between those components.
- Both principal applications rebuilt
- iOS and Android apps rebuilt
- Azure Databricks Medallion lakehouse
- The existing 227-question methodology translated into checked calculation pipelines
- A dimensional model with fifteen dimensions and eight fact tables
- Executive analytics connected to production data
- Context-aware AI Coach using aggregated analytics
- Identity, deployment and environment controls revised
The outcome
Both applications and the iOS and Android apps entered production on the revised architecture. Dashboard queries reported at four to five seconds returned in under 600 milliseconds after an unnecessary middleware layer was removed. Those figures describe this query path and implementation, not a performance guarantee for another system.
Explore Fix & OptimiseTalk through a similar situation
Tell us what you are building, buying or trying to fix. A couple of sentences is enough.