The Cloud Engineer Roadmap for 2026: AWS, Azure, or GCP?

Published on BirJob.com · March 2026 · by Ismat

The Job Posting That Made Me Rethink Everything

In early 2024, a friend in Baku forwarded me a job posting from a European logistics company opening a remote engineering hub in Azerbaijan. The title was "Cloud Engineer." The salary was $45,000/year — roughly 3x the average tech salary in Baku at the time. The requirements seemed reasonable: Linux, networking basics, AWS or Azure experience, some scripting. He asked if he should apply.

I said yes. He didn't get it. Neither did the next 40 applicants. The company eventually hired someone from Poland. When I asked the hiring manager (through a mutual contact) why they couldn't find someone locally, the answer was blunt: "Plenty of applicants knew the theory. None of them could actually set up a VPC, configure IAM properly, or troubleshoot a load balancer in a live environment. They had certifications but hadn't built anything."

That conversation stuck with me. And it mirrors what I see globally: a massive, well-documented cloud skills gap that companies keep complaining about, combined with a flood of certified professionals who can't actually do the work. The HashiCorp 2024 State of Cloud Strategy Survey found that 86% of organizations say cloud skills gaps are their biggest challenge — not budget, not technology, but people.

This roadmap is my attempt to bridge that gap. Not with another "get 5 AWS certs and you'll be fine" plan, but with an honest, practical, hands-on path that actually prepares you to do the job. Because I've watched thousands of cloud-related job postings flow through BirJob, and the pattern is clear: employers want builders, not test-takers.

The Numbers First: Why Cloud Engineering Is One of the Best Bets in Tech

Before we dive into the roadmap, let's ground this in data. Because the cloud infrastructure market isn't just "growing" — it's reshaping the entire technology industry.

• Global cloud infrastructure spending hit $84 billion per quarter in Q4 2025 according to Synergy Research Group, up from roughly $63 billion per quarter in 2023. That's over $330 billion annually flowing into cloud services.
• The U.S. Bureau of Labor Statistics projects steady growth for network and computer systems administrators through 2032, with cloud-focused roles growing significantly faster than the category average. The computer network architects category (which includes cloud architects) shows 4% growth, but this dramatically understates the cloud-specific demand because the BLS categories don't fully capture the cloud engineering role.
• Glassdoor reports the median cloud engineer salary in the U.S. at $118,000. Mid-level cloud engineers with 3–5 years earn $90,000–$130,000. Senior cloud engineers and solutions architects earn $150,000–$200,000+. At FAANG-level companies, Levels.fyi shows total compensation for senior cloud/infrastructure roles exceeding $250,000.
• The ISC2 Cybersecurity and IT Workforce Study found that cloud computing skills are the most sought-after technical competency globally, outranking security, AI, and data analytics.
• In emerging markets: cloud engineers in Azerbaijan, Turkey, and Eastern Europe earn $10,000–$20,000/year locally, but $30,000–$60,000+ working remotely for international companies. Cloud skills are among the most portable in all of tech — a VPC is a VPC whether you're configuring it from Baku or Berlin.

The opportunity is real, but so is the competition. The cloud certification market has exploded. AWS alone has issued millions of certifications. The question isn't whether cloud engineering is a good career — it obviously is — but how to stand out in a market where everyone has the same badges on their LinkedIn profile. That's what this roadmap addresses.

AWS vs. Azure vs. GCP: The Market Reality

The first decision every aspiring cloud engineer faces: which provider should I learn? The answer depends on where you want to work, and the market data is clear.

According to Synergy Research Group's latest data and Canalys Q4 2025 reports:

My recommendation:

• Learn AWS if you have no strong reason to pick otherwise. It has the most jobs, the most resources, the largest community, and the broadest service catalog. It's the safe default
• Learn Azure if you're targeting enterprise companies, financial services, healthcare, or government. If a company uses Microsoft 365, Active Directory, and .NET, they almost certainly use Azure. Also, the Microsoft + OpenAI partnership has made Azure the default for companies deploying AI workloads
• Learn GCP if you're interested in data engineering, ML/AI infrastructure, or Kubernetes. GCP's BigQuery, Vertex AI, and GKE are genuinely best-in-class. Also a good choice if you're targeting Google or the companies in its orbit

Critical insight: The underlying concepts are 90% transferable. VPCs, load balancers, IAM policies, object storage, compute instances, container orchestration — the patterns are identical across providers. The service names differ. Learn one deeply, and switching to another takes weeks, not months. For a detailed breakdown of every certification, see our Cloud Certifications Ranked article.

The Roadmap: 12 Months to Cloud Engineer

This roadmap assumes you have basic programming skills (any language) and are comfortable using a command line. If you've never opened a terminal, spend two weeks on Linux basics first. I've broken this into four phases, each building on the last.

Phase 1: Foundations (Months 1–3)

Goal: Understand how the internet works, how Linux works, and how cloud providers organize their services. This is the foundation everything else sits on.

Linux Fundamentals (Weeks 1–4)

Cloud runs on Linux. Over 95% of cloud workloads run on Linux. If you're coming from a Windows-only background, this is your highest priority.

• Command line basics: navigation (cd, ls, pwd), file manipulation (cp, mv, rm, chmod, chown), text processing (grep, sed, awk, cat, less)
• Users and permissions: users, groups, file permissions (rwx), sudo, /etc/passwd, /etc/shadow
• Process management: ps, top, htop, kill, systemctl, journalctl
• Package management: apt (Debian/Ubuntu), yum/dnf (RHEL/CentOS)
• Shell scripting: Bash basics — variables, loops, conditionals, functions. Every cloud engineer writes Bash scripts
• SSH: key-based authentication, config files, tunneling. You'll SSH into servers daily

Best resource: Linux Upskill Challenge (free, community-driven, excellent). For more depth: Google IT Support Certificate covers Linux fundamentals well.

Networking (Weeks 3–8)

Networking is where most aspiring cloud engineers struggle, and it's where most cloud misconfigurations happen. You cannot skip this.

• TCP/IP model: layers, what happens when you type a URL
• DNS: how domain names resolve, A records, CNAME, NS records, TTL. You'll configure DNS constantly
• HTTP/HTTPS: request/response cycle, status codes, headers, TLS/SSL
• IP addressing: IPv4, CIDR notation, subnetting. This is essential for designing VPCs
• Firewalls and security groups: inbound/outbound rules, port ranges, stateful vs stateless
• Load balancing: L4 vs L7, health checks, target groups, ALB vs NLB
• VPN and private networking: how traffic flows between on-premises and cloud

Best resource: Practical Networking (YouTube) is outstanding. The Bits and Bytes of Computer Networking on Coursera (Google) for structured learning.

Cloud Provider Fundamentals (Weeks 7–12)

Now you pick your provider and learn the core services. I'll use AWS as the primary example, with Azure and GCP equivalents noted.

Hands-on requirement: Don't just read about these. Sign up for the AWS Free Tier, Azure Free Account, or GCP Free Tier and actually build things. Launch an EC2 instance. Create an S3 bucket with proper permissions. Set up a VPC with public and private subnets. Configure a load balancer. These free tiers are generous enough for months of practice.

First Certification (Weeks 10–12)

Get your foundational certification during this phase. This isn't the end goal — it's a structured forcing function to ensure you've covered the basics.

• AWS: Cloud Practitioner (CLF-C02) — $100 exam, broad but shallow
• Azure: AZ-900: Azure Fundamentals — $99 exam
• GCP: Cloud Digital Leader — $99 exam

These foundational exams are deliberately easy. That's the point. They force you to survey the entire service catalog and understand the vocabulary. Don't spend more than 2–3 weeks studying for them.

Phase 2: Core Cloud Engineering (Months 4–6)

Goal: Go from "I understand cloud concepts" to "I can design and build cloud infrastructure."

Infrastructure as Code (Weeks 13–18)

This is where cloud engineering separates from cloud operations. A cloud operator clicks through the console. A cloud engineer writes code that creates, modifies, and destroys infrastructure programmatically. In 2026, Infrastructure as Code (IaC) is non-negotiable.

Learn Terraform first. I know Pulumi is technically more elegant (real programming languages vs. HCL), and I know CloudFormation is "native" if you're on AWS. But the market has spoken. The Stack Overflow 2024 Developer Survey and the HashiCorp State of Cloud Strategy Survey consistently show Terraform as the dominant IaC tool. It appears in more job postings than all alternatives combined. It's multi-cloud. It has the largest ecosystem of modules and providers.

What to learn in Terraform:

• Core concepts: providers, resources, data sources, variables, outputs
• State management: remote state (S3 + DynamoDB for locking), state file structure, terraform import
• Modules: writing reusable modules, module composition, the Terraform Registry
• Workspaces and environments: managing dev/staging/prod
• Best practices: DRY code, naming conventions, tagging strategies, Terragrunt for large codebases

Best resource: HashiCorp's official Terraform tutorials are excellent. Build real infrastructure, not just read about it.

IAM Deep Dive (Weeks 16–18)

If there's one thing that separates competent cloud engineers from dangerous ones, it's IAM (Identity and Access Management). This is where security lives, and getting it wrong has real consequences.

• Principle of least privilege: give each entity exactly the permissions it needs, nothing more
• Policies and roles: how to write IAM policies, role assumption, cross-account access
• Service-linked roles: how cloud services authenticate with each other
• Identity federation: SSO, SAML, OIDC — how enterprise users access cloud resources
• Common mistakes: wildcard permissions ("*"), overly permissive security groups, exposed access keys in code

Associate-Level Certification (Weeks 22–26)

Now get your associate certification. This one actually matters for getting interviews.

• AWS: Solutions Architect Associate (SAA-C03) — the most popular cloud certification in the world. $150 exam
• Azure: AZ-104: Azure Administrator Associate — $165 exam
• GCP: Associate Cloud Engineer — $200 exam

Study approach: Don't just memorize service names. For each service, ask yourself: when would I use this vs. the alternative? What are the cost implications? What are the failure modes? The best exam prep combines a course (like those from Adrian Cantrill or Stephane Maarek on Udemy) with hands-on lab time.

Phase 3: Containers, Orchestration, and Serverless (Months 7–9)

Goal: Master containerization and understand when to use containers vs. serverless vs. traditional compute.

Docker (Weeks 27–30)

If you don't know Docker in 2026, you're not a cloud engineer. Full stop. Containers are the packaging format that the entire industry has standardized on. Every CI/CD pipeline builds containers. Every microservice runs in a container. Every cloud provider's container service uses Docker images.

• Core concepts: images, containers, Dockerfiles, layers, volumes, networks
• Dockerfile best practices: multi-stage builds, minimizing layers, using specific tags, non-root users, .dockerignore
• Docker Compose: multi-container applications, service dependencies, local development environments
• Container registries: ECR (AWS), ACR (Azure), Artifact Registry (GCP), Docker Hub
• Security: scanning images for vulnerabilities, signing images, secrets management in containers

Best resource: Docker's official Get Started guide. Then containerize a real application — not "hello world," but something with a database, a backend, and a frontend. Build, push to a registry, and deploy it to your cloud provider.

Kubernetes (Weeks 31–38)

Kubernetes is the container orchestration platform that won the orchestration wars. It's complex, powerful, and increasingly table-stakes for mid-to-senior cloud roles. The managed Kubernetes services — EKS (AWS), AKS (Azure), GKE (GCP) — handle the control plane, but you still need to understand how it works.

Best resource: Kubernetes official tutorials, then Mumshad Mannambeth's CKA course on Udemy. For hands-on practice: KillerCoda (free interactive Kubernetes labs).

Serverless (Weeks 35–38)

Not everything needs Kubernetes. Serverless is the right choice for event-driven workloads, APIs with variable traffic, and data processing pipelines. Knowing when to use serverless vs. containers is as important as knowing how.

• AWS Lambda / Azure Functions / Google Cloud Functions: Write functions that respond to events (HTTP, queue messages, file uploads, schedules)
• API Gateway: Exposing Lambda functions as REST/HTTP APIs
• Event-driven architecture: SQS/SNS (AWS), Event Grid (Azure), Pub/Sub (GCP)
• Serverless limitations: cold starts, execution time limits, memory limits, debugging difficulty, vendor lock-in
• When to use serverless vs. containers: serverless for spiky/unpredictable traffic and simple event processing; containers for steady-state workloads, complex applications, and when you need full control

Phase 4: Production Skills and Job Readiness (Months 10–12)

Goal: Build the production-grade skills that separate junior from mid-level engineers.

CI/CD Pipelines (Weeks 39–42)

A cloud engineer who can't set up a CI/CD pipeline is like a developer who can't use Git. You need to automate the build, test, and deploy process.

• GitHub Actions: the most popular CI/CD platform for open-source and startups. Learn workflows, jobs, steps, secrets, environments
• GitLab CI: popular in enterprise. YAML-based pipeline definitions
• Provider-native: AWS CodePipeline, Azure DevOps Pipelines, Google Cloud Build
• ArgoCD: GitOps-based continuous delivery for Kubernetes. Increasingly expected for K8s-heavy shops

Monitoring, Logging, and Observability (Weeks 41–44)

• The three pillars: metrics (Prometheus, CloudWatch), logs (ELK/OpenSearch, CloudWatch Logs), traces (Jaeger, X-Ray)
• Dashboarding: Grafana is the industry standard. Learn to build actionable dashboards, not vanity metrics
• Alerting: PagerDuty, OpsGenie, or native cloud alerting. On-call rotation management
• Cost monitoring: cloud costs are a real engineering concern. Learn Infracost for Terraform cost estimation, native cost explorers, and tagging strategies for cost allocation

Security (Weeks 43–46)

Security isn't a separate phase — it should be embedded throughout. But there are cloud-specific security topics you need to master:

• Network security: security groups, NACLs, WAF, DDoS protection
• Data encryption: at rest (KMS), in transit (TLS), key rotation
• Secrets management: AWS Secrets Manager, Azure Key Vault, HashiCorp Vault
• Compliance: SOC 2, GDPR, HIPAA — understand the compliance frameworks your industry requires
• Security scanning: vulnerability scanning, configuration auditing, tfsec for Terraform security

The Portfolio Projects (Weeks 44–48)

Build 2–3 projects that demonstrate real cloud engineering skills. Here are proven project ideas that hiring managers actually respect:

1. A fully IaC-managed three-tier web application: VPC with public/private subnets, load balancer, auto-scaling group, RDS database, S3 for static assets. All deployed with Terraform. CI/CD pipeline for deployments. This single project demonstrates 80% of what a junior cloud engineer does
2. A Kubernetes-based microservices deployment: 2–3 services running on EKS/AKS/GKE, Helm charts, Ingress, horizontal pod autoscaling, Prometheus + Grafana monitoring. Shows you can handle container orchestration
3. A serverless data pipeline: Events trigger Lambda/Cloud Functions, data flows through a queue, gets processed and stored. Include error handling, dead-letter queues, and monitoring. Shows event-driven architecture skills

Put all of these on GitHub with detailed READMEs, architecture diagrams, and a section explaining the design decisions. A hiring manager should be able to read your README and understand why you chose each service and configuration, not just what you deployed.

The Certification Path: A Strategic Approach

I've already covered individual certifications in depth in our Cloud Certifications Ranked article. Here's the strategic sequence I recommend:

The rule: Never get more than one certification ahead of your hands-on experience. A certification without matching skills is worse than useless — it sets expectations you can't meet in an interview. Build first, certify second.

Career Progression: Where Cloud Engineering Takes You

Lateral moves:

• Cloud Engineer → DevOps/SRE: Natural transition. Add CI/CD depth, SLO frameworks, and incident management. Read our DevOps vs. SRE vs. Platform Engineer comparison
• Cloud Engineer → Security Engineer: Specialize in cloud security. Huge demand, premium pay. The cybersecurity careers guide covers this path
• Cloud Engineer → Platform Engineer: Build internal developer platforms. The Gartner prediction that 80% of large orgs will have platform teams by 2026 is tracking true
• Cloud Engineer → Solutions Architect (vendor side): Work for AWS, Azure, or GCP helping customers adopt cloud. Often the highest-paying path short of management

Free Tier Resources for Hands-On Practice

You do not need to spend money to practice cloud engineering. Every major provider has a generous free tier, and there are dozens of free practice platforms. Here's your toolkit:

Critical warning: Set billing alerts on every cloud account. Set them at $1, $5, $10, and $50. I have personally seen people accidentally run up $3,000+ bills on "free tier" accounts because they left a large EC2 instance running or forgot to delete a NAT gateway. Cloud providers are not going to warn you before charging your credit card.

The AI Elephant in the Room

Every tech career article in 2026 needs to address AI, and cloud engineering is no exception. But the story here is different from most roles. AI isn't threatening to replace cloud engineers — it's creating more demand for them.

Here's why:

• AI workloads need infrastructure: Every LLM, every ML model, every AI application runs on cloud infrastructure. Someone needs to provision the GPU clusters, manage the storage, configure the networking, and keep the systems running. That someone is a cloud engineer
• AI-generated code needs infrastructure: As AI tools help developers write more code faster, more applications get built. More applications means more infrastructure. More infrastructure means more cloud engineering work
• AI is not good at infrastructure: Cloud infrastructure is stateful, has real-world consequences (downtime, security breaches, cost overruns), and requires understanding complex interactions between dozens of services. AI assistants can help write Terraform modules and suggest configurations, but they can't architect a multi-region disaster recovery strategy that accounts for your specific compliance requirements, latency constraints, and cost budget

What AI will change about cloud engineering:

• Faster IaC writing: GitHub Copilot is already good at generating Terraform boilerplate. You'll write IaC faster, but you still need to know what you're generating
• Automated troubleshooting: AI-powered observability tools will help diagnose issues faster. Datadog and Dynatrace are building AI-driven root cause analysis. You'll spend less time grepping logs and more time fixing the underlying issues
• Cloud cost optimization: AI tools will automatically identify waste and suggest right-sizing. But someone still needs to understand the trade-offs and make decisions
• Higher baseline expectations: If AI handles the routine tasks, junior cloud engineers will be expected to operate at a higher level faster. The floor rises, but so does the value of truly skilled engineers

Bottom line: Learn to use AI tools (Copilot, ChatGPT for debugging, AI-powered monitoring), but don't worry about AI replacing cloud engineering. If anything, the AI revolution is the best thing that's happened to cloud infrastructure demand since the mobile revolution.

What I Actually Think

After watching thousands of cloud and infrastructure job postings flow through BirJob, here's my unfiltered take:

Start with AWS unless you have a specific reason not to. I know Azure is growing faster in enterprise. I know GCP has superior data tools. But AWS still has the most jobs, the most resources, the largest community, and the broadest service catalog. If you're in the Azure ecosystem at work, learn Azure. If you're targeting Google, learn GCP. For everyone else, AWS is the rational default. The concepts transfer — you're not locked in forever.

Terraform is non-negotiable. I've seen too many "cloud engineers" who only know how to click through the console. That's cloud operations, not cloud engineering. If you can't codify your infrastructure, you can't version control it, review it, test it, or reproduce it. Learn Terraform deeply. It's the single skill that most consistently separates hired candidates from rejected ones in the cloud engineering interviews I've heard about.

Kubernetes is important but don't start there. I see beginners who jump straight to Kubernetes without understanding networking, Linux, or basic cloud services. That's like trying to learn React before you understand HTML. Kubernetes is a complex system that assumes you already know how containers, networking, storage, and load balancing work. Build that foundation first, or K8s will feel like dark magic instead of logical orchestration.

Certifications open doors; skills close deals. An AWS Solutions Architect Associate on your resume will get you past the initial HR screen. But the technical interview will expose whether you actually know the material or just memorized exam dumps. The best approach: study for the cert by building real things, not by doing practice exams. The cert is a side effect of genuine competence, not a substitute for it.

The cloud skills gap is real, and it's your opportunity. Companies genuinely cannot find enough qualified cloud engineers. But "qualified" means "can actually do the work," not "has certifications." The 86% of companies reporting cloud skills gaps aren't looking for more badge collectors. They're looking for people who can set up a secure, well-architected, cost-optimized cloud environment and keep it running. If you build that capability through hands-on practice, the market is extremely favorable.

Multi-cloud is a resume buzzword, not a learning strategy. Don't try to learn AWS, Azure, and GCP simultaneously. Learn one deeply. Once you're proficient, picking up a second takes weeks because the concepts are identical. Companies that claim to want "multi-cloud" experience usually just want someone who deeply understands one provider and can learn another when needed.

The Action Plan: Start This Week

Don't just bookmark this and close the tab. Here's what to do in the next 7 days:

1. Day 1: Sign up for the AWS Free Tier (or Azure / GCP). Set a billing alert at $1. Launch your first EC2 instance (or VM). SSH into it. Install nginx. See "Welcome to nginx" in your browser. Celebrate. Then terminate the instance so you don't get charged.
2. Day 2: Open a terminal. If you're on Windows, install WSL2. Run through the first 3 lessons of the Linux Upskill Challenge.
3. Day 3: Install Terraform. Follow the Get Started tutorial for your chosen provider. Write your first main.tf that creates a single resource. Run terraform plan. Read the output carefully.
4. Day 4: Install Docker. Pull the nginx image. Run a container. Map port 8080 to port 80. See the nginx page. Then write a Dockerfile for a simple application (even a static HTML page). Build it. Run it.
5. Day 5: Browse 5 cloud engineer job postings on BirJob or LinkedIn. List every skill they mention. Map each skill to a phase in this roadmap. Identify your three biggest gaps.
6. Day 6: Create a GitHub repository called "cloud-engineering-portfolio." Write a README listing 3 projects you plan to build over the next 6 months. Include architecture diagrams (even rough ones drawn in draw.io).
7. Day 7: Block 1 hour per day for cloud engineering study. Put it on your calendar as a recurring event. This is a 12-month journey, and consistency is the only thing that makes it work.

Sources

• Synergy Research Group — Q4 2025 Cloud Market Data
• Canalys — Global Cloud Services Q4 2025
• U.S. Bureau of Labor Statistics — Network and Computer Systems Administrators
• U.S. Bureau of Labor Statistics — Computer Network Architects
• Glassdoor — Cloud Engineer Salaries 2026
• Levels.fyi — Cloud Engineer Total Compensation
• HashiCorp — State of Cloud Strategy Survey 2024
• ISC2 — Cybersecurity and IT Workforce Study
• AWS Solutions Architect Associate Certification
• Azure Administrator Associate (AZ-104)
• GCP Associate Cloud Engineer Certification
• HashiCorp Terraform Associate Certification
• CNCF Certified Kubernetes Administrator (CKA)
• Terraform — Official Site
• Kubernetes — Official Tutorials
• Docker — Get Started Guide
• Linux Upskill Challenge
• DevOps Roadmap — roadmap.sh
• Adrian Cantrill — AWS Training
• KillerCoda — Interactive Cloud Labs
• Grafana — Observability Platform
• Gartner — What Is Platform Engineering

I'm Ismat, and I build BirJob — a platform that scrapes 9,000+ job listings daily from 77+ sources across Azerbaijan. If this roadmap helped, check out our other career guides: Cloud Certifications Ranked, DevOps vs. SRE vs. Platform Engineer, Best Free Certifications 2026, and The Data Analyst Roadmap.