Apleona Intelligence · Azure architecture and completion of the AI platform

Starting point and objective

Apleona had an existing AI-assisted prototype of Apleona Intelligence for internal use. I took over the organisation and leadership of further implementation at that stage. Together with the client team, the objective was to develop the application further and bring it to successful completion. The focus was on code quality and maintainability, Azure service configuration, reliable processing and the technical implementation of data protection requirements.

The assignment combined project coordination, architecture and hands-on development. The existing application and infrastructure provided the starting point. I identified which components could be retained, which changes were needed and the sequence in which we could implement them together with Apleona.

Role and responsibilities

As project lead, I structured the work and coordinated technical development with the client. As software and Azure architect, I helped set the technical direction and assessed the relationships between the web application, AI services, background processing, data storage and cloud infrastructure. I also implemented changes myself as lead developer and in the DevOps work.

I turned findings from code and architecture reviews into specific GitHub issues. These included clear problem descriptions, affected components, implementation steps and acceptance criteria. I consolidated existing and new tasks, removed overlaps and made dependencies between work packages visible. This provided a traceable basis for joint implementation and subsequent development by the client team.

Approach and hands-on implementation

I began by reviewing the existing architecture, application code and infrastructure definitions. I examined both individual component behaviour and the way components worked together. This covered API routes, chat and search features, AI integrations, asynchronous document processing and Azure resource configuration. I used the findings to distinguish immediately actionable improvements from longer-term architecture and operational measures.

I then reviewed and revised code and configurations and implemented prioritised changes myself. An initial implementation package comprised 29 individually documented changes covering Terraform and infrastructure, CI/CD, application code, AI agents and Azure AI Foundry, background processing and documentation. The changes were submitted for joint review in a pull request with a test plan and traceable individual commits.

A second package added eleven further implementations. In the application code, I adapted 18 API routes to the intended use of NextRequest. Other changes addressed workspace membership limits and clearer handling of temporary search unavailability. This work covered specific features and failure cases within the existing application.

In Azure and DevOps, I worked on infrastructure definitions and configurations. This included integrating Azure Key Vault to manage an Application Insights configuration, Cosmos DB diagnostics and a retention policy for container images. The first implementation package also included changes to CI/CD, AI processing and background processing.

I actively incorporated data protection requirements into the technical work. I identified relevant GDPR concerns, documented the required action and implemented specific safeguards. These included reducing personal data in telemetry and traces, filters for personal entities and defined retention periods for selected datasets. These measures were coordinated with the other application and infrastructure changes.

Collaboration, handover and results

Implementation was carried out jointly with Apleona's internal team. I combined requirements discussions and prioritisation with hands-on development, presenting decisions in a way that led to concrete next steps. GitHub issues, pull requests and supporting documentation made the changes traceable.

At completion, two documented implementation packages and a prioritised roadmap for further development and operations were available. The handover covered the implemented changes, their context and recommendations for areas such as monitoring, resilience, resource management and access rights. This enabled the client team to distinguish completed work from subsequent actions.

The client team carried out the internal production rollouts. My responsibilities covered organising and leading the joint implementation, architecture, hands-on development and technical handover. Together, we successfully completed Apleona Intelligence to the client's full satisfaction.

Technologies

Microsoft Azure, Azure Container Apps, Azure AI Foundry, Azure Cosmos DB, Azure Key Vault, Application Insights, Azure Container Registry, Terraform, GitHub, GitHub Actions, CI/CD, Next.js and TypeScript.