The main convenience of it is when deriving formulas, rather than when applying them. With all the constants set to 1, they don't need to be tracked throughout the formula, and can be put back in at the very end by looking at the units. Sprinkle the appropriate (hbar*c) or (mu0*epsilon0) at the end, and then you get your constants back.

Honestly I would expect something like that to be happening at a hardware level. The number of bits in a memory address for the ground sensing radar is very interesting. Or the algorithm to determine vehicle acceleration given the voltage reading of a solid-state sensor vs the baseline. The metric system vs the imperial system is not an interesting distinction in these contexts.