GitOps MLOps Platform on AKS
A production-shaped reference platform that takes ML services from commit to an observable AKS workload using infrastructure as code and pull-based delivery.
Explore architecture ↗CLOUD · DEVOPS · MLOPS · LLMOPS · AGENTOPS
Hands-on engineering labs for resilient Azure infrastructure, repeatable delivery and observable machine learning systems.
$ terraform apply --auto-approve
✓ Azure network ready
✓ AKS cluster provisioned
$ argocd app sync ml-platform
✓ 6/6 workloads available
✓ telemetry pipeline active
$
SELECTED WORK
Each lab is built as a reusable system: automated, documented and designed around real operating concerns.
A production-shaped reference platform that takes ML services from commit to an observable AKS workload using infrastructure as code and pull-based delivery.
Explore architecture ↗The same platform outcome implemented with Azure Repos, Pipelines, Artifacts and Boards—a practical comparison of delivery operating models.
View lab outline ↗Metrics, dashboards, distributed traces and centralized logs wired into Kubernetes to make model and platform behavior visible.
Read implementation notes ↗FLAGSHIP ARCHITECTURE · CASE 01
A GitOps delivery path separates build from deployment. GitHub Actions validates and packages; Argo CD continuously reconciles the cluster.
Application, model & IaC
→Test, scan & plan
→Build & publish image
→Argo CD syncs desired state
→Metrics, logs & traces
DESIGN PRINCIPLE Every deployment is traceable, repeatable and recoverable from Git.
CAPABILITY MAP
From network foundations to model telemetry, the work connects the pieces teams depend on in production.
Azure networking, identity, AKS, ingress, private connectivity
Terraform, GitHub Actions, Azure Pipelines, Argo CD, GitOps
Docker, Kubernetes, Helm, scaling, workload reliability
Model services, automated pipelines, metrics, logs and traces
LAB DOCUMENTATION
Every project is organized around reproducible milestones, design decisions and proof that the platform works.
Browse all repositories ↗Resource groups, VNet and subnet design, identities, state management and the AKS baseline.
Test gates, image builds, scanning, registry flow and immutable versioning.
Application manifests, Argo CD reconciliation, promotion and recovery paths.
OpenTelemetry instrumentation, Prometheus collection, Grafana views and operational signals.