Learning backend engineering can become a long list of technologies: Node.js, databases, containers, cloud services, queues, and deployment tools. The list is useful, but it is not a learning strategy by itself.
I make progress faster when each topic is attached to something I can build, break, explain, and improve. A project gives the technology a reason to exist.
Build around a real constraint
A useful project has a boundary. It might need to generate a file, persist structured data, serve multiple users, or recover from invalid input. Those constraints create better questions than a tutorial can provide.