Most interviews open with fundamentals — what CI/CD is, how containers differ from virtual machines, how Kubernetes schedules pods, and what Terraform does — before moving into troubleshooting scenarios like a failed deployment or a disk-full server. The answers that stand out pair a clear definition with a real production example, because interviewers want proof you have operated systems, not just studied them. This scenario-heavy format applies even to remote and international roles, so prepare four or five hands-on stories you can adapt to different questions.

At five years, questions shift from definitions to design and ownership. Expect to architect a full CI/CD pipeline, plan a highly available Kubernetes setup, handle a production outage, optimize cloud costs, and justify trade-offs like managed services versus self-hosted tools. Interviewers also probe leadership — mentoring juniors, leading incidents, and improving on-call processes. Prepare detailed stories backed by numbers, such as deploy time reduced or downtime avoided, instead of listing tools.

Fresher-level interviews focus on foundations: Linux commands, Git branching and merge conflicts, what CI/CD is, writing a simple Dockerfile, basic Kubernetes objects, core networking like DNS and HTTP, and navigating one cloud console. What separates successful freshers is proof of hands-on work, so build a small home lab, containerize a simple app, and push a working pipeline to GitHub — then be ready to explain every decision behind it.

Free PDFs are scattered across GitHub repositories and study sites, but most are recycled theory lists that have not kept pace with how interviews are actually conducted today, where scenario and troubleshooting questions dominate. If you prefer a downloadable format, Robin, a CKA-certified DevOps engineer, offers structured question-and-answer sets on Topmate compiled from real interviews, including CI/CD Q&As with diagrams and real-world cloud Q&As, so you practice at the same depth you will face in the room.

GitHub is a strong starting point — popular repositories aggregate hundreds of community-shared questions, which is ideal for building a topic checklist. The catch is quality: answers are uneven, rarely include diagrams, and seldom explain the reasoning behind a design decision. Use GitHub lists to identify what to study, then practice answering each question out loud with your own examples rather than memorizing someone else's wording.

The most frequent ones cover pods versus nodes, how the kubelet and API server interact, Services and Ingress, ConfigMaps and Secrets, liveness and readiness probes, rolling updates, and the role of etcd. For senior rounds, add RBAC, NetworkPolicies, Helm, autoscaling, and live debugging of a CrashLoopBackOff pod. Strong answers explain not just what each component does, but what breaks when it fails.

Entry-level questions test whether you know objects and commands; experienced rounds test judgment. Expect multi-cluster design, GitOps workflows, capacity and cost planning, security hardening, zero-downtime cluster upgrades, and war stories from production incidents. You will often be handed a broken manifest or a failing deployment and asked to talk through your diagnosis step by step, so practice reasoning aloud instead of reciting facts.

Typical scenarios include a pod stuck in CrashLoopBackOff, a node going NotReady, a rolling update failing halfway, a PVC that will not bind, and a service that is unreachable inside the cluster. Spin up a local cluster with minikube or kind, break things deliberately, and fix them — that muscle memory is exactly what interviewers test. Robin also offers a set of 25 staff-level Kubernetes Q&As with diagrams built from these real scenarios on Topmate.

Prioritize the fundamentals — what infrastructure as code is, terraform init, plan, and apply, the state file and why it is sensitive, remote backends with state locking, providers, modules, workspaces, and fixing drift. Then go one layer deeper with importing existing resources, handling secrets, and managing multiple environments. Interviewers can tell the difference between someone who has run Terraform and someone who has only read about it, so get hands-on with a real project first.

These usually revolve around state and team workflows — a colleague's apply overwrote your changes, the state file got corrupted, two engineers ran apply at once, or infrastructure drifted from the code. A strong answer follows a pattern: identify the root cause, name the exact commands or workflow you would use, such as state locking, import, or a remote backend, and mention preventive guardrails like plan reviews in CI. Practice narrating that structure before the interview.

Expect questions on the core services — Azure Repos, Boards, Pipelines, and Artifacts — plus YAML pipeline structure with stages, jobs, and tasks, classic versus YAML pipelines, Microsoft-hosted versus self-hosted agents, service connections, variable groups, and approval gates. Deeper rounds add migration scenarios like moving from Jenkins, pipeline security, and deploying to AKS. Be ready to sketch an end-to-end pipeline for a sample app on a whiteboard.

Core areas include image versus container, how Dockerfile layering and caching work, multi-stage builds, CMD versus ENTRYPOINT, volumes and bind mounts, networking modes, Docker Compose, and registry workflows. Security questions are increasingly common too — running containers as non-root, scanning images, and keeping image sizes small. Interviewers often ask you to write or debug a Dockerfile live, so practice one for a simple app until it is second nature.

Go beyond tool syntax: know the stages of a healthy pipeline — build, test, scan, deploy — how artifacts are versioned and promoted, what triggers a run, and how rollbacks work. Be able to compare deployment strategies like blue-green and canary, and explain where secrets fit safely. The highest-signal exercise is drawing a complete pipeline from commit to production and defending each stage, and Robin's Topmate store includes a dedicated set of 25 CI/CD interview Q&As with diagrams for exactly this.

Revise the areas DevOps teams actually live in: file permissions and ownership, systemd services, process management and kill signals, checking disk and memory usage, journalctl and log locations, SSH and networking basics with ss and curl, and text tools like grep, awk, and sed. Classic troubleshooting prompts include a server that is out of disk space or a service that refuses to start — practice diagnosing both on a Linux VM until the commands come automatically.