Cloud & DevOps
Linux, AWS, containers, infrastructure as code and the observability that tells you whether any of it is working. Built around one service you take all the way to production.
Finish this and you can stand up a containerised service on AWS from a Terraform definition, ship it through a pipeline, and answer the only question that matters at 3am: what changed, and how do you know?
-
Linux and networking
BeginnerThe shell, the filesystem, processes and the network stack. Every layer above this is an abstraction over these, and every outage eventually drops you back down to them.
- The Missing Semester of Your CS Education Shell, scripting, editors, version control and debugging — the tooling nobody teaches and everybody is assumed to know. Start here.
- Linux Journey Short, structured lessons from the filesystem up to networking. Good for filling gaps you did not know you had.
- Cloudflare Learning Center DNS, TLS, CDNs and load balancing explained clearly and without a sales pitch. Read the DNS and TLS sections in full.
- The Linux Documentation Project — guides Dated in places, still the clearest free reference on processes, permissions and the boot sequence.
Build this
A hardened server you built by hand
Take the cheapest VM you can rent. Create a non-root user, lock SSH to keys, configure the firewall, install a web server, and get a real certificate on a real domain.
Then write the runbook: every command, in order, so a stranger could repeat it. That runbook is the thing Terraform will replace two topics from now — and you will understand exactly what it is replacing.
Done when: A reachable HTTPS site and a runbook precise enough for someone else to follow.
-
Git and pipelines
BeginnerBranching you can explain, and a pipeline that makes the tests non-optional.
- Pro Git Chapters 2, 3 and 7. Chapter 10 — Git Internals — is what turns rebase from frightening into obvious.
- Learn Git Branching A visual sandbox for the operations everyone is afraid of. Do the remote levels too.
- GitHub Actions documentation Focus on reusable workflows, caching and OIDC to your cloud account. Long-lived cloud keys in CI are the credential leak everyone regrets.
-
AWS essentials
IntermediateThe dozen services almost every architecture actually uses — and the IAM model that governs all of them.
- AWS Skill Builder AWS's own training. The free Cloud Practitioner track is the fastest route to a working vocabulary.
- AWS Well-Architected Framework Read the Security and Reliability pillars properly. These are the questions a cloud interview is drawn from.
- IAM security best practices Least privilege, roles over users, no long-lived keys. Almost every published AWS breach traces back to ignoring one of these.
- Amazon VPC — User Guide Subnets, route tables, NAT and security groups. Build one by hand in the console once, then never again by hand.
Build this
A budget alarm, before anything else
Ten minutes of work that saves the single most common beginner disaster. Set a budget, wire an alarm to your email, and tag every resource you create from here on.
Then leave a small workload running for a week and reconcile the bill against what you expected. Cost is an engineering constraint, and reading a bill is a skill.
Done when: A working budget alarm and a one-page reconciliation of a week's real spend.
-
Docker and Kubernetes
IntermediatePackage once, run anywhere — then the orchestrator, honestly assessed. Most teams need containers; far fewer need Kubernetes.
- Docker — Get started Through to multi-stage builds. A 1.2GB image where a 90MB one would do is the most common first mistake.
- Building best practices for Dockerfiles Layer caching, non-root users, and why your build takes four minutes when it should take twenty seconds.
- Kubernetes Basics The interactive tutorial, in a browser, with no cluster to install. Do this before deciding whether you need Kubernetes at all.
- Kubernetes — Concepts Workloads, services and configuration first. Skip the rest until something forces you into it.
Build this
Containerise a real application and run it
Take an app you have already built. Write a multi-stage Dockerfile, get the image under 200MB, run it as a non-root user, and deploy it to a managed container service.
Record the image size and cold start before and after your optimisations. Numbers make it a portfolio piece rather than an anecdote.
Done when: A published image, a running service, and a before/after table of size and start time.
-
Infrastructure as code
IntermediateThe console is for looking. Terraform is for building — reviewably, repeatably, and with a record of who changed what.
- Terraform tutorials Do the AWS "Get Started" track against your own account, not a sandbox. Real credentials change how carefully you read a plan.
- Terraform — State Remote state and locking, before you ever share a workspace with another person. Corrupted state is the worst afternoon in this discipline.
- AWS — Infrastructure as code Useful context for why the practice exists, and where CloudFormation and CDK fit beside Terraform.
Build this
Rebuild your hand-made server in Terraform
Take the runbook from the first topic and turn it into a Terraform definition: network, instance or container service, security groups, DNS, certificate.
Then destroy everything and apply it again from nothing. If the second environment is not identical, your definition is incomplete — find out where.
Done when: A repository that can create and destroy the whole environment twice, identically.
-
Observability and reliability
AdvancedMetrics, logs and traces — and the operational discipline that decides what to do when they turn red.
- Site Reliability Engineering Free in full. Chapters on service level objectives, monitoring and postmortem culture are the ones that change how you work.
- Prometheus — Overview The data model and PromQL. Instrument one service properly rather than ten badly.
- Grafana tutorials A dashboard nobody looks at is decoration. Build one that answers a specific question you actually have.
- OpenTelemetry documentation The vendor-neutral standard for traces and metrics. Worth knowing before you get locked into anyone's agent.
Build this
An SLO, an alert, and a postmortem you wrote
Define a service level objective for your deployed service, instrument it, and alert on the error budget rather than on raw CPU.
Then break it on purpose — kill the database, saturate the CPU — and write a blameless postmortem: timeline, detection, mitigation, and what you would change.
Almost nobody entering this field arrives with a written postmortem. It is the strongest single artefact you can bring to a DevOps interview.
Done when: An SLO document, a working alert, and a postmortem of a failure you caused deliberately.
Not sure this is your path?
A path is only worth the hours if it is the right one. The counselling session tests the direction you arrived with rather than simply agreeing with it.
Find your directionClose, but not quite yours?
These nine are fixed on purpose — complete, sequenced, and ready today. If your direction sits between two of them, we will build the path around you instead.
Ask for a custom path