Microsoft Azure

Microsoft Azure without surprises

Azure has many building blocks, and it is easy to end up with resources nobody quite knows the reason for. We build solutions where every resource has a clear owner and a clear purpose, and where the whole environment can be recreated from code. We have worked on Azure platforms with many environments and services, from Azure Functions and App Service to API Management, Front Door, and private networking.

Security is part of the foundation. Services authenticate with Managed Identity, secrets live in Key Vault, and traffic flows through private endpoints and WAF where needed.

Azure Functions, Container Apps, or App Service — how we choose

The choice of hosting affects cost, scaling, and how much the team has to operate itself. Our rule of thumb:

• Azure Functions for event-driven jobs, scheduled integrations, and workflows with Durable Functions. A good fit when load varies a lot and executions are short.
• App Service for APIs and web apps with steady traffic. Deployment slots give zero-downtime releases, and the operating model is familiar to most teams.
• Container Apps when the service is already a container, needs sidecars or several processes, or should scale to zero without being rewritten as Functions.
• Static Web Apps for frontends that do not need their own server.

We rarely recommend AKS for smaller platforms. Kubernetes offers great flexibility, but also an operational burden that only pays off when many teams share the same cluster. Whatever the choice, we use Managed Identity between services, so connection strings with secrets do not need to live in configuration.

Bicep and Terraform across many environments

As a platform grows, it becomes important that test, staging, and production are set up the same way. We have worked with versioned Bicep module libraries using template specs, where teams reuse approved building blocks instead of copying configuration, and with declarative YAML models that compile to Bicep, where each environment is described briefly and the rest is derived automatically.

For event-driven solutions we use Event Hubs, Event Grid, and SignalR Service, and for data Azure SQL, Cosmos DB, and Azure Cache for Redis. Where AI is relevant, we connect services to models in Azure AI Foundry within the same networking and access boundaries.