Interview Questions for a DevOps Engineer

Walk me through a CI/CD pipeline you've built, from commit to production.

What to look for: Build, test, artifact, deploy stages; gates and rollbacks; environment promotion; and what makes the pipeline trustworthy and fast.

What's the difference between blue-green, canary, and rolling deployments, and when would you use each?

What to look for: Understands risk trade-offs, rollback speed, traffic shifting, and the infrastructure cost of each strategy.

How do you manage infrastructure as code, and how do you prevent configuration drift?

What to look for: Declarative tooling like Terraform, state management, code review for infra, and detecting drift rather than manual console changes.

How do you manage secrets so they never end up in a repo or a container image?

What to look for: A secrets manager or vault, injection at runtime, rotation, least privilege, and scanning to catch accidental commits.

What do you monitor to know a service is healthy, beyond 'is it up?'

What to look for: The golden signals (latency, traffic, errors, saturation), SLOs and error budgets, and actionable alerts over noisy ones.

How would you debug a Kubernetes pod stuck in CrashLoopBackOff?

What to look for: Reads logs and events, checks resource limits, liveness probes, image and config issues, and reasons methodically rather than restarting blindly.

How do you make a deployment safely reversible?

What to look for: Immutable artifacts, versioned releases, backward-compatible migrations, feature flags, and a tested rollback path.

How do you design autoscaling so a service handles spikes without over-provisioning?

What to look for: Scales on the right signal (CPU, queue depth, request latency), sets sane min and max, and accounts for warm-up time and cost.

Tell me about a major outage you were part of. How did you handle it and what changed after?

What to look for: Calm under pressure, clear communication, a blameless post-mortem, and concrete prevention like alerts, runbooks, or automation.

Describe a manual process you automated. What was the impact?

What to look for: Bias toward eliminating toil, measuring time saved or errors reduced, and improving reliability not just convenience.

How do you balance shipping fast with keeping production stable?

What to look for: Uses guardrails like progressive delivery and error budgets to enable speed safely rather than treating them as opposites.

Tell me about a time you introduced a tool or practice that developers initially resisted.

What to look for: Builds trust, leads with developer experience, demonstrates value with a pilot, and brings the team along rather than mandating.

Deploys are passing CI but failing intermittently in production. How do you investigate?

What to look for: Compares environments, checks for config and data differences, race conditions, and resource limits, and adds observability.

Cloud costs doubled this month with no traffic change. How do you find the cause?

What to look for: Uses cost tooling and tags, finds the spiking resource, checks for orphaned infra or misconfigured autoscaling, and fixes it.

A teammate wants to hotfix straight to production, bypassing the pipeline. How do you respond?

What to look for: Protects the pipeline as the source of truth, finds a safe expedited path, and avoids one-off changes that cause drift.

An alert is paging the on-call constantly but nothing is actually broken. How do you fix the alerting?

What to look for: Tunes thresholds to symptoms users feel, ties alerts to SLOs, removes noisy non-actionable alerts, and prevents alert fatigue.