DevOps

DevOps interviews test whether you can package, deploy, operate, and recover services—not whether you memorized every kubectl flag. Strong answers connect containers, orchestration, infrastructure as code, and runtime operations to team delivery and incident constraints.

This domain spans Docker (images through Compose), Kubernetes (workloads through Helm), infrastructure as code (Pulumi, Terraform, state, environments), and runtime operations (health checks, probes, logs, metrics, alerts, incidents).

On interviews: for a Node.js API on Kubernetes, sketch image build, deploy rollout, config/secrets, health probes, observability signals, and how you would mitigate a bad deploy.

Common pitfalls: mutable `latest` images; compose as production orchestration; namespaces without policies; paging on log noise; IaC state in Git.

The trade-off is shipping speed versus reproducibility, security boundaries, and operability under failure.

Checklist:

• Promote immutable artifacts across environments.
• Separate build, deploy, and runtime configuration.
• Design health and observability before incidents.
• Use child topics for stack-specific depth.