NFA minimization without determinization

I believe the problem is just known as NFA minimisation as well. It's NP-hard in general, I believe.

One fruitful approach may be Bisimulation Minimization [Paige, Tarjan 1987], and subsequent derivatives.

This presentation has some notes on the tractability of the problem as well as references to some approaches with their relative merits spelled out.

There's an implementation of the Kameda-Weiner algorithm for NFA minimization here: https://github.com/coder0xff/parlex_legacy/blob/132e4a23a599140d22b18ead832626f0c607340f/Automata/NFA.cs#L641