[ptero]

My 2c: there isn't really a lot of math underneath practicing many ML jobs. But I would read up on basics. Get a basic understanding of optimization and gradient descent. If you are not math inclined, do not bother with untangling chain rules for networks, but play with a few toy, non ML, low dimensional problems. That understanding will help a lot.

And I would look at applying existing models to business problems. That I think has a lot of demand now. Solve a few problems for yourself, make a few blog posts and apply.