Advanced Azure: AKS, DevOps, and AI

Advanced Azure: AKS, DevOps, and AI

Azure Kubernetes Service manages Kubernetes control-plane operations while customers govern workloads, nodes, networking, identity, policy, and upgrades. Azure DevOps and GitHub Actions automate delivery. Microsoft Foundry is the current name for Microsoft's unified Azure AI platform, succeeding Azure AI Foundry.

Core ideas

• AKS integrates with Microsoft Entra ID, Azure RBAC, managed identities, networking, registries, and Azure Monitor.
• Azure Pipelines and GitHub Actions can authenticate through workload identity federation.
• Bicep is Azure's declarative language over Azure Resource Manager resources.
• Microsoft Foundry organizes models, agents, tools, evaluations, observability, and governed project access.

Design and operating model

Choose AKS only when Kubernetes capabilities justify platform complexity. Build immutable artifacts, store them in a controlled registry, deploy through reviewed pipelines, and separate platform from application responsibilities. For AI, govern data, model selection, evaluation, content safety, identity, network access, monitoring, and rollback.

Example

apiVersion: apps/v1
kind: Deployment
metadata:
  name: learning-api
spec:
  replicas: 3
  selector:
    matchLabels: { app: learning-api }
  template:
    metadata:
      labels: { app: learning-api }
    spec:
      containers:
        - name: api
          image: example.azurecr.io/learning-api@sha256:REPLACE
          resources:
            requests: { cpu: 100m, memory: 128Mi }
            limits: { memory: 256Mi }

Run examples in a disposable training subscription. Replace names, Regions, identifiers, scopes, sizes, and policies with reviewed values, and confirm current Azure CLI and service requirements before production use.

Production guidance

• Use workload identity, private networking, Azure Policy, maintained node images, and controlled cluster administration.
• Version delivery templates and avoid pipeline credentials with subscription-wide standing access.
• Evaluate AI systems for quality, safety, latency, cost, privacy, drift, and responsible use before production.

The chapter quiz follows the lesson and checks both service vocabulary and architecture decisions.