[Jugurtha]

We have colleagues who are similar to you, PhD. What we ended up doing is building an internal machine learning platform to reduce the number of "taps on the shoulder". They had trouble with setting up the environments, dealing with libraries, systems dependencies, etc. In addition to that, they relied on others to get data, fix their environment, deploy their models, or showcase their work to clients.

It wasn't optimal because we were having bottlenecks and variance: some people could move through the stack and do it all, but you either had them or you had to train them and it took time.

- [0]: https://iko.ai