The paper describes an experiment aimed at creating a categorical and interactive stratification schema for the population a major Russian city (St. Petersburg). We used the data on friendship ties of 3200 adults to create a network of ties among occupations. We then used the Louvaine community detection algorithm to identify six clusters. The clusterization obtained distinguished between skilled manual, routine non-manual and professional occupations demonstrating that close social ties are more likely to be found within, rather than between, their boundaries. However, in contrast to Goldthorpe’s class schema, the algorithm also identified cleavages between sectors of professional occupations (pedagogical/ artistic, clerical, etc.) The boundaries between such groups of occupations are reproduced inter-generationally even in the absence of considerable economic inequality between them. We demonstrate that clusters of occupations differ in their lifestyles and consumption patterns (e.g. consumption of highbrow culture) even controlling for age, gender, and education. We interpret the clusterization as evidence of the existence of milieus confined within institutional barriers of social sectors. Such milieus, rather than classes, serve as the building blocks of social structure defined through intensity of interaction or lifestyles.