Antibodies currently constitute the most rapidly growing class of human therapeutics. One of the major problems encountered in antibody-based therapies is that these antibodies tend to aggregate under high concentration formulations required for disease treatment. The aggregation in turn leads to a decrease in antibody activity and could elicit an immunological response. In this study, we employ molecular simulation tools in collaboration with experimental techniques to understand the mechanism behind antibody aggregation. Based on this mechanistic understanding, we engineer antibodies with enhanced stability. Using detailed atomistic molecular dynamics simulations of the antibody Fab, Fc fragments and that of the full antibody we show that there are specific regions on the surface of antibody that are prone to aggregation. These aggregation prone regions are either always surface exposed, or exposed due to dynamic fluctuations or conformational changes observed in our molecular simulations. Based on these simulations, we develop a parameter that indicates the location and size of these aggregation prone regions. Furthermore, we perform mutations in these specific aggregation prone regions to engineer antibodies with enhanced stability.