Cloud (AWS / Azure)

Cloud engineering is where architectural decisions turn into monthly invoices. The skill worth developing is not memorising every AWS service name — there are hundreds, and most of them do not matter for most workloads — but knowing which primitives (compute, storage, network, identity, messaging) exist on every major cloud and when each is the right choice. This track focuses on the small set of services you will actually use for 90% of backend work, and on the discipline of designing for cost and blast radius, not just for the happy path.

What this track covers

The core AWS and Azure primitives — compute (EC2, ECS Fargate, Lambda / App Service, Container Apps), storage (S3, RDS, DynamoDB / Blob, SQL Database, Cosmos DB), networking (VPCs, security groups, load balancers), and identity (IAM, RBAC). Then the practical topics: deploying a Spring Boot or FastAPI service, choosing between serverless and containers, cost estimation, and multi-AZ design for reliability.

Who this is for

Backend engineers whose service is about to leave a laptop for a real environment for the first time, and mid-level engineers moving into a cloud-heavy role at a company committed to AWS or Azure.

Prerequisites

Comfortable with Linux, SSH and basic networking (subnets, ports, DNS)
Have built at least one service you now need to host somewhere real
Basic Docker — most cloud deployment paths assume you can ship a container image
A credit card for the free tier, and the discipline to set a billing alarm before you touch anything

Recommended learning order

// beginner

Start with 'Deploying Spring Boot to AWS' (or the equivalent path on Azure). Get one service exposed on a public URL behind a managed load balancer, backed by a managed database. Set a billing alarm before you go further.

// intermediate

Study identity and networking properly — IAM roles vs users, VPC design, security groups vs NACLs. Move your database credentials to a secrets manager. Add CloudWatch or Azure Monitor dashboards for the basics: request rate, error rate, latency, database connections.

// advanced

Multi-region design, disaster recovery (RTO / RPO), cost optimisation (reserved instances, spot capacity, right-sizing), and infrastructure-as-code with Terraform or CDK so the environment is reproducible.

Common mistakes to avoid

Long-lived access keys in a .env file instead of IAM roles / managed identities
Public S3 buckets 'just for a test' that end up in a data-leak headline
Provisioning the biggest instance type because 'we might need it' — cloud bills scale badly
No billing alarm, discovering a runaway workload weeks later
Treating one region as 'good enough' for a service that has an SLA requiring more

Career relevance

Cloud competence is now assumed for essentially every backend role above junior. The specific cloud matters less than the ability to reason about identity, networking and cost — those concepts transfer between AWS, Azure and GCP with a rename.

Featured tutorials in this track

Cloud (AWS / Azure)7 min

Serverless Java: Spring Boot on AWS Lambda with GraalVM

Deploy Spring Boot on AWS Lambda with GraalVM native image to eliminate cold starts — including build configuration, API Gateway integration, benchmarks and cost comparisons vs containers.