As more applications and commerce systems moved beyond managed platforms, I learned to deploy and troubleshoot the infrastructure underneath them.
Owning the failure instead of forwarding the ticket
Self-hosting changes the nature of a problem. A blank page might be application code, an environment variable, a container, a port, a reverse proxy, DNS, SSL or caching. Learning infrastructure meant learning to follow the request through the entire stack.
That ability is useful even when a managed platform is involved because it creates a clearer mental model of what is actually happening in production.
Headless commerce made the layers visible
Work with MedusaJS, Node, PostgreSQL and headless architecture made the separation between storefront, backend, database, search and deployment explicit. Instead of one ecommerce admin containing everything, each layer could be chosen and operated independently.
That flexibility also created more responsibility — exactly the kind of tradeoff that has appeared throughout my journey from marketplaces to owned infrastructure.