I'm thinking that writing support software for scientific research at universities could be like that (partly because they can't just afford to hire teams to do the job, so a single guy often must suffice). Do you have any other ideas?
I develop ERP/MRP integrations for a manufacturing company. So we have an ERP system which we purchased and I build things on top of it to enhance processes within the company. Production trackers, product configurators, issue monitoring, things to speed up engineering management etc. I'm sure there are other roles like this in manufacturing firms because very few of these ERP systems actually do exactly what you want out of the box.
It's good because I'm not just a code monkey. I (and others) identify issues/improvements, I design the solution, I implement the solution and then I train people on how to use it. There's a lot of autonomy.