| I worked there relatively recently. DigitalOcean is organizationally incapable of making anything in-house anymore. I want to blame leadership - and to be clear, leadership sucks - but the problems are pervasive through every layer of the organization. The only significant launches in the last 4-5 years have all been acquisitions or built by partner companies and whitelabeled. Every system and every team has massively circular dependencies on one another , so it's just a massive circle of "we can't move until they move". The tech debt is insane. Everything is slowed down by terribly-run and massively underfunded internal "platforms" teams for kubernetes, CI/CD, various internal databases, etc. If you want to build something useful you basically have to ignore upper management and do it in secret until it's done and so integral to the systems that they have no choice but to allow you to support it. Asking leadership outright to invest in minor maintenance for systems the entire company depends on is never approved. The bar for engineering practices, code quality, and system design quality is comically low. All the systems are massively distributed, but there is no understanding of distributed systems issues. I was told multiple times that "CAP theorem doesn't apply here" and gaslit that an asynchronously replicated MySQL instance that sometimes spiked to multi-minute replication lag to the read replicas should just be used as if it were completely consistent between master and read replicas. Tons of stuff was just run as singletons with hand-rolled in-memory rate limiting to avoid having to understand distributed locking or semaphores. These systems inevitably start falling over a few months after creation, but you're not allowed to evolve it into a correct system, you just have to support that garbage forever. Brain drain everywhere due to low salaries. Even engineers barely capable of committing working code were getting fat raises to leave. |