A sculptor with a traditional background would probably tell you to sculpt directly into pose, specially if it’s going to be just a still or for 3D printing. Is faster and a much straightforward approach.
But, if you sculpt in a T pose first you can take advantage of mirror sculpting for the main forms and then pose for the final details of the sculpt. The problem with this is that you have to take the extra steps in making a simple rig for the character first. And, if you sculpt with Dyntopo rigging can be a real PITA.
Zbrush’s transpose works great for this kind of workflow, the Speedsculpt addon offers something similar inside Blender, check it out.
I think you should try both methods and see which one feels and works better for you.