Blog Post
Build these top cloud projects to gain hands-on experience with AWS and Azure. Learn how to deploy virtual machines, use IaC, and set up CI/CD pipelines.
Checking sign-in status...
A GitHub repository with three documented cloud projects beats a certification with no evidence. Cloud engineering roles in 2026 screen for shipped infrastructure, not tutorial completions.
The cloud platform you choose matters less than what you build. AWS, Azure, and GCP all expose the same core concepts: storage, compute, networking, managed databases, and automation.
Most engineers get stuck in passive learning. They watch tutorials, understand the theory, and still freeze when asked to explain a real architecture in an interview. The gap is application.
These five projects are sequenced from easiest to hardest. Each one builds on the infrastructure from the previous project. Start with Project 1. Do not skip ahead.
| Order | Project | Core Skill | Time | Difficulty |
|---|---|---|---|---|
| 1 | Static Resume Hosting | Storage and DNS | 2-3 days | Beginner |
| 2 | VM Web Server | Compute and networking | 1 week | Beginner |
| 3 | Two-Tier Application | Managed databases and IAM | 1-2 weeks | Intermediate |
| 4 | Infrastructure as Code | Terraform or Bicep | 2-3 weeks | Intermediate |
| 5 | CI/CD Pipeline | Deployment automation | 3-5 days | Advanced |
Two rules. First, do not skip ahead. Second, break your deployments on purpose. If everything works on the first try, you are not learning.
Cost warning: Most of these projects fit within free tiers, but free tiers have limits. AWS gives 750 hours of t2.micro or t3.micro EC2 per month for 12 months. Azure gives $200 credit for 30 days plus 12 months of free services. GCP gives $300 credit for 90 days. Set a billing alert on day one. Delete resources you are not actively using.
Time: 2-3 days.
What you build: A basic HTML resume hosted on AWS S3 or Azure Blob Storage. The design does not matter. The goal is to understand how the cloud serves static files.
What you learn: How cloud storage works. The difference between public and private access. DNS and custom domain setup.
Why this project matters: Static hosting is the simplest possible cloud deployment. It teaches the storage layer before you touch compute.
How to start:
index.html file.Common failure point: The bucket policy syntax. If your site returns a 403 error, the bucket is private and the policy is missing or malformed. Read the error message carefully before searching for a fix.
Time: 1 week.
What you build: A Linux or Windows VM running a web server (Nginx, Apache, or IIS) that serves a simple page.
What you learn: Firewalls and security groups. Public and private IPs. Port access for HTTP (80) and HTTPS (443). SSH or RDP access.
Why this project matters: This is the most common interview scenario. "A user says the site is down. Walk me through how you would debug it." You will debug it yourself during this project.
How to start:
/var/www/html.Common failure point: The security group. Your server is running, your web server is installed, but the page will not load. Nine times out of ten, the security group does not allow inbound traffic on port 80. This is the lesson. The application is fine. The network is blocking you.
Time: 1-2 weeks.
What you build: A web frontend connected to a backend database. Use a managed service like Amazon RDS or Azure SQL. Do not install the database on a VM.
What you learn: Secure service-to-service communication. Connection strings and credentials. Keeping databases private with no public exposure. IAM roles for service access.
Why this project matters: This is where architecture decisions start to matter. You will choose between managed and self-hosted, public and private, hardcoded and rotated credentials. Every choice has a trade-off.
How to start:
Common failure point: Connection timeouts. The web server cannot reach the database. The cause is usually a security group rule that does not reference the correct source. Security groups can reference other security groups, not just IP ranges. This is the pattern that keeps databases private.
Time: 2-3 weeks.
What you build: Delete the environment from Project 3 and recreate it entirely using Terraform or Azure Bicep.
What you learn: Automation. Repeatability. Real-world deployment practices. Version-controlled infrastructure.
Why this project matters: IaC is a core industry skill. Job descriptions list Terraform in a majority of cloud engineering roles. If you can only click in a console, you cannot be trusted with production infrastructure.
How to start:
terraform plan to preview changes.terraform apply to deploy.terraform destroy to tear everything down.Common failure point: State management. Terraform tracks what it deployed in a state file. If you lose the state file, Terraform does not know what exists. For local projects, commit the state file to a private repository. For production, use a remote backend like S3 with locking.
Time: 3-5 days.
What you build: Automate the deployment of your static resume from Project 1. Every commit to GitHub should update the live site automatically.
What you learn: CI/CD fundamentals. Version control workflows. Deployment automation.
Why this project matters: Modern cloud workflows are automated. A developer who ships through a pipeline is worth more than one who deploys manually. This is the workflow every engineering team expects.
How to start:
.github/workflows/.Common failure point: Authentication. The pipeline needs credentials to deploy. Do not commit access keys to the repository. Use OIDC federation (GitHub to AWS) or a service principal secret stored in GitHub Secrets.
By completing these five projects, you will have hands-on experience with storage, compute, networking, managed databases, IaC, and deployment automation. You will be able to draw the architecture of what you built and explain the trade-offs behind each decision.
That is what interviewers probe for. Certifications prove you studied. Projects prove you can build.
Set a billing alert first. Do this before you create any resource. A forgotten VM can cost more than the exam fee in a week.
Delete resources when you stop working. The free tier is generous only if you use it correctly. Stop instances. Delete unused databases. Empty storage buckets.
Document as you go. Keep a README in each project repository. Include architecture diagrams, screenshots, and notes on what broke.
Use the cloud provider's own documentation. AWS, Azure, and GCP all publish step-by-step tutorials for every service used here. Follow them instead of third-party guides that may be outdated.
Do not worry about which platform. AWS, Azure, and GCP all teach the same concepts. Pick one and go deep. Adding a second platform later is easy because the concepts transfer.
Any of the three major providers works. AWS has the largest free tier for EC2 and S3. Azure integrates well if you already use Microsoft tools. GCP has the simplest networking model for beginners. Pick the one your target employers use and go deep.
Most of these fit within free tiers. AWS gives 750 hours of t3.micro EC2 per month for 12 months, plus 5 GB of S3 storage. Azure gives $200 credit for 30 days. GCP gives $300 credit for 90 days. Set a billing alert and delete unused resources. Most beginners finish all five projects for under $10.
IaC makes deployments repeatable and reviewable. Console clicks cannot be version-controlled or peer-reviewed. Terraform and Bicep let you tear down and rebuild environments without manual steps. That is how production infrastructure is managed.
The projects cover the same core services tested in AWS Certified Cloud Practitioner, Azure Fundamentals (AZ-900), and AWS Solutions Architect Associate. Building the services yourself makes the exam questions easier because you have seen how they behave in practice. Use ExamOS to test your knowledge after each project.
Pick Project 1 this weekend. Create a free-tier cloud account, write a basic HTML file, and deploy it. Momentum beats perfection. ExamOS tracks your progress with Daily Sprints and a Readiness Score, so you always know what to build next.