The extant dyadic majority-minority framework cannot adequately account for residential segregation in increasingly multi-racial America, let alone complex multi-ethnic societies in the Global South. Using well-established metrics, we develop a multi-group, multi-scalar, and multi-cleavage measurement framework and apply it to describe residential segregation in rapidly urbanizing India. We also introduce a new micro-dataset containing independent India’s first census-scale enumeration and coding of caste and religion categories (𝑛 ≈ 60 million, and ≈ 700 ascriptive groups). We find that segregation patterns in contemporary urban India represent a palimpsest of three distinct layers—near-perfect segregation as a constitutive feature of India’s traditional caste society, the colonial Indian town as the ground zero of modern racial segregation, and the segregation engendered by political economy careers of caste and religion in post-colonial India. Each layer represents a “boundary maintenance” process (à la Fredrik Barth) that accounts for the stability of caste and religion boundaries across the urbanization spectrum, ranging from small semi-urban settlements to a globalized metropolis of over thirteen million residents. Our analysis also shows how the Global South is a generative site for developing new frameworks to study segregation and is not merely an empirical site for applying methods developed in the Global North.