I see it this way (and I'm open to correction): if science wants to place the burden of proof on the person making the positive assertion, and simultaneously administer empiricism as the rational standard, then it should uphold that same standard to prove what it denies -- deity. Thus, if a scientist wants to deny the existence of any form of deity then one of two things need to happen:
1. S/he must admit his/her denial of deity is an honest leap of faith in the absence of evidence;
2. Provide evidence for the lack of a (or all) deity.
But I think you'll agree with me when I say that neither of those two options is really satisfactory. Not only because we agree that scientific empiricism is not the only way to know something, but because the sword cuts both ways: theists cannot hold to faith while simultaneously administering dogmatisms as the rational standard, and all the while deny the use of empiricism as a form of proof.
The conversation has to move past the bifurcation of "nonoverlapping magisteria" (Stephen J. Gould) and into a more wholistic appraisal of reality. Neither science nor religion needs a magisterium; we need useful conversation, and helpful perspectives from each other in the reality that we both share.