Opscraft
Platform engineering & Backstage · Lesson 1 of 5 · 3 min read

What platform engineering is

Many DevOps roles are now titled platform engineer, and senior interviews increasingly ask how you'd build an internal developer platform rather than how you'd write one pipeline. This topic explains what platform engineering is and why it emerged, how golden paths and self-service infrastructure work, how Backstage provides a catalogue, templates and docs, and how to run a platform as a product that developers actually choose to use, including how to measure it.

Why it emerged

"You build it, you run it" gave teams ownership, but it also handed every product team Kubernetes manifests, Terraform, CI pipelines, observability, security scanning and cloud permissions. Each team solved the same problems differently, onboarding took weeks, and developers spent much of their time on infrastructure instead of features. This is a cognitive load problem.

Platform engineering answers it by having a dedicated team build an internal developer platform (IDP): a curated set of tools, templates, services and automation that lets product teams build, deploy and run software through self-service, with good defaults built in. The platform team doesn't do the work for other teams (that would be a ticket queue); it builds the products that let them do it themselves.

Enlarge diagramEvery team does everythingWith a platformapp codeKubernetes manifests and Terraformpipelines, scanning, secretsdashboards and alertsapp code (the team)golden-path templates (platform)self-service infra with guardrails (platform)observability by default (platform)
Figure 1. Cognitive load: without a platform each team solves every layer; with one, teams focus on their product and consume the rest as self-service.
Swipe sideways to see all of it, or tap to open it full screen.

Team Topologies

The book Team Topologies gives the vocabulary interviewers often use:

Team type Role
Stream-aligned team Owns a product or service end to end; most teams
Platform team Provides internal services that stream-aligned teams consume as self-service, reducing their cognitive load
Enabling team Coaches teams on new skills for a while (for example, observability practices), then moves on
Complicated-subsystem team Owns a part needing deep specialist knowledge (a video codec, a pricing engine)

The preferred interaction between a platform team and its users is X-as-a-service: clear interfaces, documentation and self-service, rather than constant collaboration on every request.

What a platform includes

A typical platform for a Kubernetes-on-AWS company covers:

  • Golden-path templates for new services: repository, CI pipeline, Dockerfile, Helm chart or Kustomize base, dashboards and alerts.
  • CI/CD: reusable workflows or shared libraries, artifact and image repositories, GitOps delivery.
  • Runtime: shared EKS clusters with namespaces per team, ingress or Gateway API, autoscaling, secrets integration.
  • Self-service infrastructure: databases, queues, buckets and DNS on request, through Terraform modules or Kubernetes-style APIs.
  • Observability by default: metrics, logs and traces collected without each team setting it up.
  • Security guardrails: policy as code, image scanning and signing, least-privilege identities, secrets management.
  • A developer portal that ties it together: service catalogue, docs, templates, ownership.
Enlarge diagramdeveloper portalcatalogue, templates, docs, scorecardsgolden pathsservice templates, CIself-service inframodules, compositionsguardrailspolicy, scanning, identityCI/CD and GitOpsruntime: EKSobservabilitycloud: AWS accounts, networking, IAM
Figure 2. An internal developer platform: a portal on top of golden paths and self-service, on top of shared infrastructure.
Swipe sideways to see all of it, or tap to open it full screen.