The DevOps Roadmap: what to learn, in what order

Choose a role and test your starting point before scheduling study. The four stages below are an editorial learning sequence, not industry levels or a promise about time to promotion.

Learn processes, signals, memory limits and how to inspect Linux state. Exit 137 indicates SIGKILL under the usual shell convention; investigate cgroup events and kernel logs before calling it an OOM kill. Practise DNS resolution, routing and TCP connection diagnosis.

Add Git, one scripting language and one cloud's core services. Checkpoint: deploy a small service, reproduce a failure, collect evidence and recover it. Use documentation as you would at work.

Study Kubernetes reconciliation and scheduling if the target role uses Kubernetes. Work through infrastructure state, remote locking, drift and a reviewed change using Terraform or OpenTofu.

Build an immutable artifact and promote it through environments. Add tests, least-privilege deployment credentials and observable rollout checks. Checkpoint: demonstrate a failed canary, explain whether rollback is safe and calculate an error budget with stated units.

Practise capacity planning and failure recovery across zones or regions. Explain the consistency and cost trade-offs of your design. Include an incident exercise where you coordinate communication while another person investigates.

For current roles, extend the exercise with workload identity and software supply-chain verification. Add GPU scheduling and inference capacity when the job includes AI workloads. Checkpoint: defend a design against a new capacity constraint and revise it when the numbers fail.

Technical depth remains necessary. Add migration sequencing, adoption across teams and a budget that compares build, buy and continued operation. Describe how you would limit risk when different teams move at different times.

Checkpoint: defend a proposed infrastructure investment with costs, alternatives, milestones and evidence that would make you stop it. The exercise supports preparation; it does not substitute for the experience a specific role requires.

Measure progress by demonstrations and feedback. The time required depends on your starting knowledge and access to realistic work; this guide cannot establish a universal months-to-senior schedule.

Use supervised incident practice, recovery drills and code review. On-call experience can help, but an unsupported overnight rotation is not a training plan.