here you go: https://arxiv.org/pdf/1009.1166.pdf
The paper I was referring to used the category Poly, and literally just stated that Poly could be used for database migrations without elaborating further.
Poly is an incredibly rich category which can be used to describe a ton of stuff such as finite state machines, neural networks, game interactions, wiring diagrams and dynamical systems more generally. I wouldn't be surprised at all if Poly can describe database migrations.
One thing that bothers me about applied category theory is that this (being able to describe stuff) always seems to be sufficient for people to claim it is applied. But if the people doing the applying don't find that framing particularly useful, then to me it does not count.