The “contact hypothesis,” or the idea that effective interpersonal contact can ameliorate outgroup prejudice, is a central (even if sometimes implicit) motivation for the quantitative measurement of residential segregation. However, extant segregation metrics typically do not account for group diversity or spatial arrangements within administratively defined neighborhoods, such as census blocks, which are the sites of everyday outgroup interactions that modulate anomie or trust between groups. We introduce ONE (Outgroup Neighbor Exposure), a simple framework combining neighborhood diversity and neighbor homophily to better account for salient outgroup contact opportunities. Using a unique census-scale demographic microdataset, we apply our framework to examine outgroup contact opportunities for two of the most spatially and socio-economically marginalized groups in India—Dalits (historically “untouchable” caste groups) and Muslims. We uncover a non-monotonic association between neighborhood diversity and neighbor homophily, such that the most diverse neighborhoods are associated with the greatest hyperlocal segregation.