AWS Backend & DevOps Learning Roadmap
Amazon Web Services (AWS) is a collection of on-demand cloud computing services β compute, storage, databases, networking, and automation β that you rent instead of buying and racking physical servers. Instead of provisioning a machine, plugging in a cable, and installing an OS, you request an EC2 instance and it exists seconds later, billed by the second. This doc walks through the 24 AWS and DevOps services a backend engineer actually needs, in the order you're likely to need them, with a diagram, a "where to use it" guide, and step-by-step implementation instructions for every single one.
Use the sidebar to jump straight to any service β each one opens its own page with a full explanation, a diagram of exactly how it connects to the services around it, and the concrete steps to actually stand it up. If you're starting from zero, work through Phase 1 first: nothing else on this roadmap makes sense without IAM, a VPC, EC2, and a database already in place.
This is the shape most production backends converge on. DNS resolves to a CDN, which forwards uncached requests through a load balancer into a fleet of containerized API instances, which talk to a database, a cache, and a queue. Everything underneath is what makes that top row possible β every box below is its own page in this sidebar.
All 24 services in this roadmap
Summary & Patterns
Once the individual services make sense on their own, these four references tie them together: the order to learn them in, how urgently each one matters, the full production shape, and the single pattern worth internalizing above all others.
Recommended Learning Sequence
Recommended Priority
Key Backend Pattern β Decoupling Slow Work
The single most useful pattern on this whole roadmap: never make a client wait on work it doesn't need to wait on.
The API's only job is to validate the request, drop a message on the queue, and respond β step β happens in milliseconds, before the real work even starts. The worker pulls from the queue on its own schedule (β‘β’) and only then calls out to a slow external service (β£). If that external service is down or slow, the client never notices β it already got its response.
Official Documentation Reference URLs
Every service and tool covered in this roadmap, with a direct link to its official docs.