This is interesting because I specifically look for the opposite in interviews and design reviews. Yes, sometimes complexity is needed to handle scaling, extensibility, observability, security, and privacy, but most of the time it doesn't. I have new engineers who jump to designing big complicated modular systems with multiple levels of abstraction and I have to tell them to slow down and try something simpler. YAGNI is a real consideration.

My principal is: you don't have to do it right now, but make sure you could do it if you need to. This mainly comes down to avoiding assumptions. Simplifying assumptions can make things easier, but designs that don't leave room for extension make it really difficult to add those things later. It's a difficult balancing act, but I greatly appreciate any engineer who recognizes it and can manage it.