Most cloud roadmaps are a screenshot of a service catalogue — 200 logos, no sequence, no time budget. That is how people spend eight months learning service names and still fail the first technical round. This roadmap is ordered, time-boxed against a full-time job, and calibrated to what Indian employers actually screen for. It also tells you what to skip, which is the part nobody writes.
The Honest Prerequisites
Time: 1.5–2 hours a day, six days a week, for 6–8 months from an IT-adjacent start (support, sysadmin, QA, development), or 9–12 months from scratch. Mindset: most of this is debugging, not watching. If a broken system annoys you rather than interests you, read whether infrastructure work is really for you before spending the year.
Months 1–2: Linux and Networking (Where Most Candidates Die)
Cloud is Linux and networking wearing a web console. Skip these and every later month becomes memorisation instead of understanding — which is exactly why so many certified candidates freeze when an interviewer asks why a subnet can't reach the internet.
- Daily-drive Linux (WSL2 or a VM): files, permissions, processes, systemd, logs, SSH with keys
- Networking for real: IP and CIDR (you must be able to subnet by hand), DNS, HTTP(S), TCP, ports, NAT, routing tables, firewalls
- Bash to the level of writing a log-parsing script unaided
- Milestone: your first GitHub repo — a script that health-checks a service and alerts
CIDR and routing come back in every VPC question on the exam and in every "why can't this instance reach the database" interview scenario. Two weeks here saves two months later.
Month 3: Git and Python
Git beyond add-commit-push: branching, rebase vs merge, resolving conflicts calmly, pull-request workflow. Then Python for automation — file handling, calling APIs, boto3 basics. The single most common gap in rejected candidates is scripting: "knows the console, can't code" gets filtered fast, and it's the reason cloud engineers plateau at ₹8 LPA.
Months 4–5: One Cloud, Deep (Not Three, Shallow)
Pick one. Multi-cloud on a fresher resume reads as depth in none. AWS has the most Indian job volume; Azure dominates GCCs and enterprises; GCP leads in data and ML teams — the full decision framework is in AWS vs Azure vs GCP.
- IAM properly — roles, policies, trust, least privilege. This is 30% of the certification exam and 100% of real security incidents
- Networking in the cloud: VPC, subnets, route tables, security groups, NAT gateways, load balancers
- Compute and storage: EC2, S3 (and its storage classes), RDS, autoscaling
- Monitoring and cost: CloudWatch, billing alarms on day one — a forgotten NAT gateway is an expensive lesson everyone learns exactly once
- Milestone project: deploy a three-tier application by hand. Then break it deliberately and fix it. That is the interview story
- Certification window: this is where AWS Solutions Architect Associate fits — we've written a full 45-day SAA plan, and the real cost in ₹ (Azure is roughly a third the price)
Month 6: Infrastructure as Code
Terraform, and the discipline that comes with it: recreate everything you built by hand in month four as code — state, modules, variables — and then never touch the console again for provisioning. Every GCC interview now has an IaC round, and "I click in the console" is a disqualifying answer at anything above junior level.
Month 7: Containers and CI/CD
- Docker beyond the tutorial: lean images, multi-stage builds, debugging a container that won't start
- CI/CD with GitHub Actions: build → test → scan → deploy, with proper secrets handling
- Kubernetes fundamentals — pods, deployments, services, ingress — on a managed cluster (EKS/AKS). Depth here is what separates cloud engineers from the higher DevOps bands
- Milestone: your month-four app now builds and deploys itself on every push, provisioned entirely by Terraform
Month 8: Security, Observability and the 2026 Edge
- Cloud security as a discipline: least privilege everywhere, secret management, image scanning in the pipeline. Consider AZ-500 or AWS Security Specialty — cloud security is one of the scarcest, best-paid profiles in India
- Observability: Prometheus and Grafana, structured logging, and being able to say what you'd alert on and why
- The differentiator: basic familiarity with running GPU workloads and serving an LLM. Teams are hiring cloud engineers specifically to run AI infrastructure, and almost no candidates can speak to it
What to Skip
- A second cloud. It comes with your second job, not your first — and it costs about 20% of the effort of the first once concepts transfer
- Certification collecting. One associate cert is the sensible ceiling before employment. Three reads as breadth without depth
- Service-mesh, operators, custom controllers. Real skills, wrong layer for a first role. Learn them on the job
- Obscure services. Nobody will ask you about the 200th AWS service. They will ask why your subnet has no route to the internet
The Whole Thing on One Screen
| Months | Focus | Milestone |
|---|---|---|
| 1–2 | Linux, networking, CIDR, Bash | Health-check script repo |
| 3 | Git workflow, Python, boto3 | Automation scripts |
| 4–5 | One cloud, deep. IAM above all | 3-tier app deployed; SAA cert |
| 6 | Terraform / IaC | Everything as code |
| 7 | Docker, GitHub Actions, K8s basics | Self-deploying app |
| 8 | Security, observability, AI infra | Monitored, scanned, documented |
| 7–9 | Job hunt (overlapping) | Interview in batches |
Don't finish learning and then start applying — overlap them from month seven. And before you accept anything, check the number against our salary calculator: cloud engineers routinely under-ask by several lakh because nobody told them what the band was.
❓ Frequently Asked Questions
How long does it take to become a cloud engineer in India?＋
From an IT-adjacent background (support, sysadmin, QA, development): 6–8 months at 1.5–2 hours daily. From a complete non-IT start: 9–12 months. Anyone promising a cloud job in 90 days from zero is selling something — the Linux and networking foundations alone take two months to internalise properly.
Do I need Linux and networking before learning AWS?＋
Yes, and skipping them is the single most common mistake. Cloud is Linux and networking wearing a web console. Without CIDR, routing, DNS and firewalls you'll memorise service names instead of understanding them — which is exactly why many certified candidates freeze when asked why a subnet can't reach the internet.
Which cloud certification should a beginner take first?＋
AWS Solutions Architect Associate if you're targeting startups and product companies, or AZ-104 if you're targeting GCCs and enterprises (and Azure is roughly a third the price in India). Take it in months 4–5, after you've deployed something real — a certificate without a deployed project still fails technical rounds.
Cloud engineer or DevOps engineer — which should I choose?＋
They overlap heavily and the roadmaps share most of their content. Cloud engineering skews toward infrastructure design and provisioning; DevOps skews toward pipelines, automation and production ownership. Cloud is the slightly easier entry point for freshers; DevOps bands run higher at senior levels. Learn cloud first, add CI/CD and Kubernetes, and both doors are open.
Founder · TrueDirectory
Firoz Ahmed is the founder of TrueDirectory, India's business and education listing platform. He writes straight-talking, research-backed guides on tech careers, courses and companies — genuine editorial recommendations, never paid rankings or sponsored placements.