There are many components to the problem, but one component is institutional bias in the hiring process itself. You assume that all candidates are given a fair shake by an impartial, unbiased set of criteria that should be ferreting out the best talent and making offers to only those people. But those supposedly impartial criteria are executed upon in interviews by fallible human beings with their own personal or institutional biases, not by impartial pattern-matching robots. Consequently, many talented female engineers are passed over for inferior male ones, not the other way around.
I have interviewed literally thousands of candidates in my tech management career, and not once have I seen an inferior female candidate picked over a superior male one because of a diversity initiative. I have, however, had to work with young male interviewers to train them to properly identify their own unconscious biases and to ensure that they are truly being as impartial as possible when interviewing candidates of any gender. And to the degree that that is impossible - well, that is why we still have diversity targets.