Aging is a complex process that affects individuals from the scale of the whole body to the cell.1 In this work, we examine the aging trajectories of over 40 cell types using 7,000 proteins measured in the blood of 60,542 people with machine learning models. Our results suggest cellular aging is heterogenous between people and individual cell-types. Aging maps to disease risk in a cell-type specific manner. The accumulation of aged cell types is linked to increased mortality risk. Circulating cellular aging signatures emerge as robust disease predictors, comparable to known risk factors.