People dealing with disorders like paranoia clearly have unusually strong biases (or other issues, in the case of schizophrenia) to deal with, but exceptional epistemic rationality consists of compensating well for your biases, not of not having any in the first place.

I am confused.

If a paranoiac has "unusually strong biases" but is exceptionally good at compensating for them, he would not be diagnosed with paranoia and would not be considered mentally ill.

My understanding of epistemic rationality is pretty simple: it is the degree to which your mental model matches the reality, full stop. It does not care how bad your biases are or how good are you at overcoming them, all that matters is the final result.

I also don't think full-blown clinical paranoia is a "bias" -- I think it is exactly a wrong picture of reality that fails epistemic rationality.