I would start with camera and 2 player movement (just movement no actual fighting).
If you want multiplayer in future do it now from very start, there are problems specific only to mp games (like replication).
So MVP for you would be: smart camera that does not madly flip when your player turns, movement that is intuitive, and multiplayer (or some basic AI support, at least circling, jumping forward and backward)
For camera prototyping use 3rd person template. Make event that randomly spawns (moves) opponent dummy.
Then make camera that smoothly react to changed distances/positions.
This is enough fun for several hours of coding.
Hint: use camera arm for positioning, and interpto for smooth value changes (like distance)