If I were starting over in 2026 — knowing everything I know from 10 years in infrastructure — this is the roadmap I’d follow. It’s different from every generic roadmap you’ve seen because it integrates AI tools from day one, not as an afterthought at the end.

The philosophy: Learn the fundamentals deeply, use AI as an accelerant throughout, specialize in one AI-adjacent area at the mid-level.

PHASE 0: MINDSET (Week 0 — Free)

Before touching a single tool, internalize this: you are learning to be an infrastructure engineer in an age of AI assistance. That means your value is not “knowing the syntax” — it’s understanding systems well enough to architect, review, and debug what AI generates. Shift your learning goal from “memorize commands” to “understand concepts so deeply you can validate AI output.”

PHASE 1: UNBREAKABLE FOUNDATIONS (Months 1–3)

Linux — The Non-Negotiable Base

  • Filesystem hierarchy, permissions, process management
  • Networking: TCP/IP, DNS, HTTP/HTTPS, subnets, routing
  • Bash scripting: loops, conditionals, text processing
  • Systemd, cron, logging with journald
💡 AI Tip: Use Warp AI terminal from day one. Let it explain commands you don’t understand. Ask it “what does this awk command actually do” — it’ll explain better than man pages.

Git and Python

  • Git: branching, rebasing, merge conflicts, hooks
  • Python: scripting level — file ops, API calls, JSON parsing
💡 AI Tip: Use GitHub Copilot for Python scripting from day one. Don’t fight it — learn to review its output. This builds your reviewer instinct early.

PHASE 2: CORE CLOUD + CONTAINERS (Months 3–6)

Docker — Deep, Not Surface

  • Dockerfile: multi-stage builds, layer caching, image optimization
  • Container networking, volumes, Docker Compose
  • Security: non-root containers, distroless images, image scanning

AWS Core

  • VPC, EC2, S3, IAM, RDS, Route53, CloudWatch
  • Deploy one real application end-to-end
💡 AI Tip: Use Amazon Q Developer for all AWS-specific questions. It understands AWS services at a depth no other AI does.
AI-Era DevOps Roadmap

PHASE 3: INFRASTRUCTURE AS CODE + CI/CD (Months 6–9)

Terraform

  • Core: resources, variables, outputs, modules, state
  • Remote state: S3 + DynamoDB locking
  • Workspaces, environments, module versioning
💡 AI Tip: This is where Copilot shines brightest. Let it generate your first module, review and improve it, ask it to add encryption or tagging. You’ll write production-grade Terraform faster than engineers with 3× your experience.

GitHub Actions CI/CD

  • Build → test → security scan → deploy pipeline
  • Secrets management, environments, deployment protection rules

PHASE 4: KUBERNETES + OBSERVABILITY (Months 9–12)

Kubernetes — With Context

  • Pods, Deployments, Services, Ingress, ConfigMaps, Secrets
  • RBAC, Network Policies, Resource limits
  • Helm charts, Kustomize, GitOps with ArgoCD

Observability Stack

  • Prometheus + Grafana: metrics, dashboards, alerts
  • Distributed tracing with Jaeger or Tempo
  • Structured logging with Loki or ELK

PHASE 5: AI SPECIALIZATION — PICK ONE (Months 12–18)

By month 12, you have the foundation. Now pick your AI-era specialization:

  • MLOps Path: MLflow → Kubeflow → model serving → drift monitoring
  • AIOps/SRE Path: Deep Datadog AI / Dynatrace → SLO engineering → chaos engineering
  • Platform Engineering Path: Backstage → Crossplane → LLM integration → AI-native IDP
  • AI Infrastructure Path: GPU orchestration → LLM serving (vLLM, TGI) → inference optimization

The One Rule That Changes Everything

Every week, take one thing you did manually and automate it. In the AI era, your automation tool is an LLM. “Write a Python script that checks if our S3 buckets have versioning enabled and sends a Slack alert for any that don’t.” Do this weekly for a year. After 52 weeks, you have 52 automation scripts, deep scripting muscle memory, and deep AI prompting skills. That portfolio is worth more than any certification.