Every service I build will die

And that's exactly the point. I'm a senior software engineer at Ontime Payments, a fintech startup enabling direct-from-salary bill payments. We've deliberately built a modular, event-driven serverless architecture, and every service within it is expected to be replaced eventually. Some won't be. But we build as if they will, and that shapes everything. The basic idea Serverless has well-known benefits: no servers to manage, no patches to apply, compute that scales automatically. But the thing that's changed how we work isn't just the managed infrastructure. It's what happens when you combine small, focused functions with event-driven communication and a philosophy that any component can be killed and replaced when required. If you're unfamiliar with how serverless infrastructure works at its most basic: a user hits an API Gateway, which triggers a Lambda function. That function does its job (say, processing a payment), returns a response to the user, and raises an event to EventBridge