A hippo is two tonnes of solid mass built like a barrel on stumps. It can run at 30 km/h -- faster than Usain Bolt -- and change direction mid-charge. In water, it doesn't swim. It runs along the bottom, pushing off and gliding, covering ground faster than you can paddle away.
"Hippos are vegetarian. Imagine if they weren't." — overheard during a nature documentary
The heaviest thing in the river is also the most agile. That's the idea behind Potamus.
Operations becomes heavy as you scale -- there's no escaping this. But it doesn't have to feel heavy. It can be a weight pulling you down, or momentum driving you forward.
The best infrastructure is the kind nobody notices
When I built the ML pipeline and automation layer of Bolt's Model Lifecycle platform, the best metric of success wasn't uptime or throughput. It was that data scientists forgot the platform existed. They just did their work. They didn't think about Docker, or orchestration, or deployment configs. They thought about models.
I wrote about the specifics across several posts -- cutting build times from 40 minutes to 5, replacing deployment tickets with self-service config, what happens to team output when you remove operational friction. The common thread: complex machinery under the hood, hidden behind an interface that lets people think about their actual job.
"Potamos" is Greek for river. Hippopotamus -- "river horse." We kept the river part. Good operations should flow like one: heavy things carried effortlessly, obstacles routed around, everything moving in the same direction without anyone having to think about the current. The current flows towards one number everyone understands.