I think that is very doable, companies used to need a team of masters / PhDs to implement the ML stuff more or less from scratch, but libraries are nowadays ver

I think that is very doable, companies used to need a team of masters / PhDs to implement the ML stuff more or less from scratch, but libraries are nowadays very advanced and easier to use. Just start learning stuff that interests you, maybe keep them in a private git repo or even write a blog about them. It is realistic to aim for a junior position, and in the first few years you'll be learning a lot! Many companies utilize AWS / Azure / Google Cloud, and they offer free tiers so that you can start using them basically for free. And with the help of virtualization you can even run a local minikube "cluster" and try running Airflow and whatnot there. Although ML engineering has a bit different focus than say data scientist. But even in those roles it is useful to know how to program, do custom web UIs etc.