A thing about hiring in startups: you aren’t *actually* subject to the statistics of the broader candidate pool because you’re looking for such a tiny slice of it.
I’ve done a lot of hiring at startups in San Francisco in the past, and I’ve managed to get 50/50 gender balance in candidates a couple of times without spending a lot on advertising, and without decreasing the quality of candidates.
You do this by developing your network to include more underrepresented minorities, learning about how to create an inclusive job posting, learning how to correct for bias in the interview process (if you’re doing whiteboard algorithms questions, you’ll loose URM candidates), and so on.
For the record, at my last startup I worked at (Rdio), my team was 30% women, 30% queer (including me), and only 50% white. We were also one of the most effective teams in the entire company. You see even better stats at npm, Inc, where men are in the minority there, and they’ve been accomplishing some impressive technical feats recently.
This kind of diversity can be done at startups by exploiting the statistics related to being a startup. If you can’t do it, then you’re failing at it, because numerous other startups have shown it can be done.