We're not talking to the same engineers. I hear about vast quantities of unmaintained code, still running, half-forgotten, across codebases in multiple languages, using ancient pre-AWS APIs
Sure, some teams might have their codebase in a really poor condition. Would be hard not to, given the size of the company.
But the core engineering stuff is high quality there. The build system that the whole company uses seems extremely well made. And I doubt that such a core thing can be done so well by a company with an overall poor engineering culture.
And the inverse seems to ring true as well. Even in large companies with overall poor engineering culture, you can find a few small pockets of teams that produce great small stuff. But they are fighting against the overall inefficiency of the company, while with Amazon, the opposite is happening, as poorly performing teams are being boosted by the core things (such as build systems) being done really well.
