DevOps Engineer

Navigation, globbing, permissions and ownership - half of production issues are permissions issues.

ps, top, signals and systemd units - finding, inspecting and managing what's actually running.

grep, sed, awk, cut and pipes - slicing logs and configs at the speed of thought.

Variables, conditionals, loops, exit codes and cron - automating the toil away is the job description.

apt/yum, sudoers and SSH keys - provisioning a fresh box into a working server.

IP addressing, ports, TCP vs UDP and DNS resolution - the layered model as a debugging checklist.

Requests, status codes, certificates and termination - what 'it's probably TLS' actually means.

Security groups, reverse proxies (nginx) and L4/L7 balancing - the boxes between users and your app.

curl, dig, ss, traceroute and tcpdump basics - evidence before hypotheses.

Branching, merging, rebasing and bisect - history as a debugging tool, not just a backup.

Config and infra in repos, changes via PR - the audit log and rollback story for free.

Images vs containers, Dockerfiles and layer caching - small images, fast builds, no snowflakes.

Multi-service local stacks with volumes, networks and health checks that mirror production shape.

Tagging strategies, vulnerability scanning and multi-stage builds - images as governed artifacts.

Build → test → scan → publish → deploy as code in the repo; fail fast, cache aggressively.

Never in git: secret stores, scoped tokens and rotation - the difference between an incident and a breach.

Rolling, blue-green and canary - and rollbacks you've actually rehearsed before you need them.

Immutable builds, semver tags and 'deploy what you tested' - reproducibility as policy.

Registries for images and packages (ECR, Artifactory, GitHub Packages) - provenance and retention as first-class concerns.

EC2, S3 and EBS - instances, object storage and the pricing model that shapes architectures.

Users, roles and least-privilege policies - the security model everything else in AWS hangs off.

Subnets, route tables, security groups and NAT - private networks you can reason about.

RDS, load balancers and autoscaling groups - letting the cloud carry the undifferentiated weight.

Lambda-style functions and managed containers (ECS/Fargate) - when no-server beats your-server, and the cold-start fine print.

Providers, resources, plan/apply and variables - declarative infrastructure with a diff you can review.

Remote state, locking and reusable modules - collaborating on infra without stepping on each other.

Ansible at concept level: idempotent server configuration when you're not fully containerised.

Pods, Deployments, Services and Ingress - the vocabulary of every k8s conversation.

ConfigMaps, Secrets and probes - wiring apps to their environment the k8s way.

kubectl describe/logs/exec and reading events - the triage loop for CrashLoopBackOff and friends.

Charts and values - packaging k8s apps so deploys are parameterised, not copy-pasted.

ArgoCD/Flux syncing clusters from git, and meshes (Istio/Linkerd) for traffic and mTLS - where platform teams go next.

Prometheus and Grafana: the four golden signals and dashboards that answer questions.

Structured, centralised logs and distributed traces - following one request across services.

Alert on symptoms users feel, page rarely, and let error budgets arbitrate speed vs stability.

Runbooks, on-call hygiene and blameless postmortems - reliability as a habit, not a hero.

Take the OneRoadmap DevOps certification to verify the fundamentals on your resume.