Skip to content
Akwannya Hub
All tech roadmaps

Cloud & Infrastructure

Cloud Engineering

Cloud engineers build and run the infrastructure applications live on. The work is choosing the right managed service, connecting things securely, and being able to say what the bill will be and why.

6 stages2 projectsBeginner friendly4 to 6 months, part time

Start here

Create one free-tier account, set a billing alarm before anything else, and work through Stage 1. You need no prior cloud experience, but you do need to be comfortable in a terminal.

01

Foundations before the cloud

The cloud is someone else's computer. These are the ideas that make the rest make sense.

Required

Linux and networking

Every cloud service is a Linux machine or a network rule wearing a friendly name. Without these, the console is just vocabulary.

What to learn

  • Linux command line
  • SSH and key pairs
  • IP addressing and subnets
  • DNS
  • HTTP and TLS in outline

Tools

  • A terminal
  • dig
  • curl

Practice

Rent the cheapest virtual machine you can find, SSH in, install a web server, and reach it from your browser.

Next: A cloud account.

02

Your first cloud account

Pick one provider and stay with it. Breadth across three at once teaches you nothing well.

Choose one provider for the next six months.

AWS has the largest market share and the most job listings. Azure dominates where organisations already run Microsoft. Google Cloud is strongest in data and Kubernetes. The concepts transfer; the menus do not.

Pick one

AWS

The most job listings globally and in Ghana, and the largest body of learning material.

What to learn

  • Regions and availability zones
  • IAM users, roles and policies
  • Billing alarms
  • The free tier and its edges

Tools

  • AWS Console
  • AWS CLI

Next: Compute and storage.

Pick one

Microsoft Azure

The default where an organisation already runs Microsoft 365 and Active Directory, which is a great many of them.

What to learn

  • Regions and resource groups
  • Entra ID and RBAC
  • Cost management
  • The free tier

Tools

  • Azure Portal
  • Azure CLI

Next: Compute and storage.

Pick one

Google Cloud

Strongest in data, analytics and Kubernetes, with the cleanest developer experience of the three.

What to learn

  • Projects and regions
  • IAM
  • Budgets and alerts
  • The free tier

Tools

  • Google Cloud Console
  • gcloud

Next: Compute and storage.

03

The core services

Four building blocks account for most of what most systems need.

Required

Compute

Where your code runs, and the first place cost gets away from people.

What to learn

  • Virtual machines
  • Containers as a service
  • Serverless functions
  • Autoscaling
  • Choosing between them honestly

Practice

Run the same small application three ways: on a VM, in a container service, and as a function. Compare cost and effort.

Next: Storage.

Required

Storage and databases

Getting storage class and database choice right is worth more than any amount of application tuning.

What to learn

  • Object storage
  • Block storage
  • Managed relational databases
  • Backups and restore
  • Storage classes and lifecycle rules

Practice

Delete a database and restore it from a backup. If you have not tested a restore, you do not have a backup.

Project

beginner

A static site with a real backend

Host a static site on object storage behind a CDN, with a serverless function writing to a managed database. Set a billing alarm.

  • Your provider
  • Its CLI

You can put something on the internet using managed services and know what it costs.

Next: Networking, which is the stage most people skip.

04

Networking and security

The stage most people skip, and the one interviewers ask about. It is also where outages come from.

Required

Cloud networking

Almost every 'it works locally but not in the cloud' problem is here.

What to learn

  • Virtual networks and subnets
  • Public versus private subnets
  • Security groups and firewall rules
  • Load balancers
  • NAT and internet gateways

First understand

  • IP addressing and subnets

Practice

Put a database in a private subnet and reach it only from an application server. Then prove it is unreachable from the internet.

Next: Identity, the other half of the same problem.

Required

Identity and secrets

Over-permissive access is the most common serious cloud misconfiguration there is, and the easiest to avoid once you understand roles.

What to learn

  • Least privilege in practice
  • Roles rather than long-lived keys
  • Secrets management
  • Audit logging
  • Encryption at rest and in transit

Practice

Give a service the narrowest policy that still lets it work. Start with nothing and add permissions until it stops failing.

Next: Writing it all down as code.

05

Infrastructure as code

Click once and you will click again. Clicking is not reproducible and cannot be reviewed.

Required

Declarative infrastructure

It turns infrastructure into something reviewable, repeatable and deletable, which is what makes experimentation safe.

What to learn

  • Declarative configuration
  • State and why it matters
  • Modules and reuse
  • Plan before apply
  • Reviewing an infrastructure change

Tools

  • Terraform or OpenTofu
  • Your provider's own tool

Practice

Rebuild your entire earlier project from a repository, then destroy it and rebuild again.

Next: Operating it.

06

Operate and prove it

The part that turns a certificate into evidence.

Required

Monitoring and cost

In the cloud, cost is a production concern. An unmonitored system with an unwatched bill is not finished.

What to learn

  • Metrics and dashboards
  • Log aggregation
  • Alerting worth waking for
  • Cost attribution and tagging
  • Right-sizing

Project

advanced

A production-style deployment

A multi-service application defined entirely in infrastructure as code: private networking, a managed database, a load balancer, monitoring, alerts and a documented monthly cost.

  • Terraform
  • Your provider
  • Its monitoring service

You have architecture written down, alerts that fire, and a number for what it costs per month.

Next: Certification, only if it helps you.

Optional

Certification

It opens doors at organisations that filter by it. It is not a substitute for having built something, and it never will be.

What to learn

  • Associate-level cloud certification for your chosen provider

Next: A specialisation.

Where this leads

You do not have to pick one now. These are the directions this path opens up once you are working.

DevOps Engineering

Automate the path from a commit to production.

View roadmap

Site Reliability Engineering

Own availability and performance as an engineering problem.

View roadmap

Solutions Architecture

Design systems for organisations rather than building them yourself.

View roadmap

Cloud Networking

Specialise in the connectivity layer.

View roadmap

Continue your journey

The paths closest to this one. Skills overlap more than the job titles suggest.

DevOps Engineering

Automate the path from a commit to something running in production.

View

Site Reliability Engineering

Treat uptime as an engineering problem with a budget, not a hope.

View

Solutions Architecture

Design systems that fit the business, the budget and the team you actually have.

View

Cloud Networking

Connect cloud systems to each other, to offices, and to the internet — safely.

View

We teach this on Cloud Build with Peers

Our programs are free, taught live, and built around the same progression. Join one and work through it alongside other people.

See our programsCreate a free account