Personal preference at this time is to compose in Cubase and Mix in Pro Tools. I like starting fresh for a mix after fine tuning the MIDI performance.
I also like keeping the sequence relatively clean and minimal. I put the reverb in Pro Tools (on a separate computer) and use it as a general verb while writing. Then start printing audio into Pro Tools (the audio is printed pre-reverb sends in PT) and you are building your mix session as you print.
I do understand the concerns voiced here about constant revisions, and the difficulty that can it can present if you are constantly printing to audio, mixing, then need revisions, back to MIDI, then print again, etc.... Especially when you want the demos to be top notch and fully mixed, but need to work quickly as well. A condensed STEM mix may help there; e.g. have a narrower set of STEMs that you export / print and mix from (rather than a full multi-track mix session). If a revision is needed, edit MIDI, export the same STEMs, put them on the same tracks in a new mix session and continue mixing from there.
As far as my personal backup process, every day I work on a piece I save...
- New version of the .cpr or .ptx file with today's date
- MIDI file with today's date
- VEP .vesp64 file with today's date
So there's a series of backups for everything each day I work on a piece, and the mix session contains a backup of the audio.
In short, it is more convenient to do it all in one session (MIDI and mix), but I feel that my production always turns out better when I separate the two tasks.