Cloud & Infrastructure
Platform Engineering
Platform engineers treat infrastructure as a product and other engineers as customers. The measure of success is not how clever the platform is, but how quickly a new developer can go from an empty repository to running code without filing a ticket.
Start here
This path assumes you already know how to deploy things. If you do not, DevOps Engineering comes first. Platform engineering is that knowledge turned into something reusable by people who do not have it.
Before you begin
- Solid with containers and orchestration
- Fluent in infrastructure as code
- Have built CI/CD pipelines other people used
- Can write production-quality code in at least one language
The product mindset
The stage that actually distinguishes this path. Everything else is DevOps skills applied differently.
RequiredPlatform as a product
Internal platforms fail for product reasons, not technical ones. They get built for the platform team's taste and then nobody adopts them.
Platform as a product
Internal platforms fail for product reasons, not technical ones. They get built for the platform team's taste and then nobody adopts them.
What to learn
- Identifying who your users are
- Interviewing engineers about their actual friction
- Measuring adoption honestly
- Choosing paved paths over mandates
- Saying no to one-off requests
Practice
Next: The technical foundation.
RequiredMeasuring developer experience
Without numbers, platform work becomes an argument about taste that the loudest person wins.
Measuring developer experience
Without numbers, platform work becomes an argument about taste that the loudest person wins.
What to learn
- Lead time for change
- Deployment frequency
- Change failure rate
- Time to restore
- Time to first successful deploy for a new joiner
Next: Foundations.
The technical foundation
You cannot abstract what you do not understand well.
RequiredKubernetes in depth
Most platforms are built on it, and platform teams are the ones who get called when its abstractions leak.
Kubernetes in depth
Most platforms are built on it, and platform teams are the ones who get called when its abstractions leak.
What to learn
- Operators and custom resources
- Admission controllers
- RBAC and namespaces as tenancy
- Networking and service mesh in outline
- Multi-tenancy and resource isolation
Tools
- Kubernetes
- Helm
- Kustomize
Next: Infrastructure as code, at scale.
RequiredInfrastructure as code at scale
One team's Terraform and fifty teams' Terraform are different problems.
Infrastructure as code at scale
One team's Terraform and fifty teams' Terraform are different problems.
What to learn
- Reusable modules
- State organisation across teams
- Policy as code
- Drift detection
- Safe automated apply
Tools
- Terraform or OpenTofu
- Open Policy Agent
- Crossplane
Next: The golden path.
The golden path
The opinionated, supported route from idea to production. This is the actual deliverable of a platform team.
RequiredTemplates and scaffolding
Every team writing its own pipeline from scratch is the waste a platform exists to remove.
Templates and scaffolding
Every team writing its own pipeline from scratch is the waste a platform exists to remove.
What to learn
- Service templates
- Pre-wired CI/CD
- Sensible defaults for logging and metrics
- Versioning templates without breaking existing services
Tools
- Cookiecutter or Backstage templates
- Reusable CI workflows
Project
intermediateA golden path template
A service template that produces a repository with a working pipeline, container build, health checks, metrics, and a deployment to a test environment. From nothing to running in under fifteen minutes.
- Your CI system
- Kubernetes
- A template engine
A new service can reach production without anyone asking the platform team a question.
Next: Self service.
RequiredSelf-service infrastructure
A platform where you still have to file a ticket for a database is a ticket queue with better branding.
Self-service infrastructure
A platform where you still have to file a ticket for a database is a ticket queue with better branding.
What to learn
- Declarative resource requests
- Guardrails rather than gates
- Automatic provisioning
- Quotas and cost attribution
Tools
- Crossplane
- Terraform automation
- Kubernetes operators
Next: The developer portal.
RecommendedDeveloper portal and catalogue
Knowing which services exist, who owns them and where their runbook is, is a real and underrated problem.
Developer portal and catalogue
Knowing which services exist, who owns them and where their runbook is, is a real and underrated problem.
What to learn
- Service catalogue
- Ownership metadata
- Documentation as part of the platform
- Discoverability
Tools
- Backstage
Next: Governance.
Guardrails and governance
Making the safe thing the easy thing, rather than making the unsafe thing forbidden.
RequiredPolicy as code
Standards enforced by review meetings do not scale. Standards enforced by the pipeline do.
Policy as code
Standards enforced by review meetings do not scale. Standards enforced by the pipeline do.
What to learn
- Admission policy
- Required labels and ownership
- Image provenance
- Automated compliance evidence
Tools
- Open Policy Agent
- Kyverno
Next: Multi-tenancy.
AdvancedMulti-tenancy and cost
Once several teams share a platform, isolation and cost attribution stop being optional.
Multi-tenancy and cost
Once several teams share a platform, isolation and cost attribution stop being optional.
What to learn
- Namespace and cluster tenancy models
- Network policy
- Resource quotas
- Chargeback and showback
Project
advancedA minimal internal platform
A platform where a developer can request a service and a database declaratively, receive a deployed environment with monitoring and enforced policy, and see what it costs. Document it as you would a product.
- Kubernetes
- Crossplane or Terraform
- OPA
- Backstage
Another engineer used your platform end to end without your help, and you have their feedback written down.
Where this leads
You do not have to pick one now. These are the directions this path opens up once you are working.
Continue your journey
The paths closest to this one. Skills overlap more than the job titles suggest.
DevOps Engineering
Automate the path from a commit to something running in production.
ViewSite Reliability Engineering
Treat uptime as an engineering problem with a budget, not a hope.
ViewCloud Engineering
From never opening a cloud console to running real workloads on one.
ViewMLOps
Make machine learning repeatable, deployable and observable.
ViewYou do not have to do this alone
Our programs are free, taught live, and built around the same progression. Join one and work through it alongside other people.