Most likely, he is not using the term ‘optimal’ in a rigorous sense. Perhaps in theory you could specify a utility function for the time it takes to perform a shuffle, and another utility function for how much randomness the shuffle has, and then describe an optimal shuffle in the sense that it maximizes your own total utility according to those two utility functions. But I don’t really know if that is practical.
Meanwhile, you could try
searching Google on the keywords shuffle and cards. One page that looks reasonable is
wikiHow article How to Shuffle and Deal Texas Holdem.