John Bell’s seminal theorem revealed that quantum correlations between space- like separated events contradict any local hidden variable explanation for many such correlations. This work lead to resolution of the Einstein-Podolsky-Rosen paradox which emerged from a belief that at the most fundamental level nature respects local-realism. Subsequent experiments confirmed that quantum correlations indeed violate local-realism. Such Bell nonlocal correlations are considered as a powerful resource which lead to many possible applications in quantum information processing. It also forms a basis to ask further foundational questions like what are the limits of quantum nonlocal correlations and how these limits can be understood better. In this talk, we discuss some foundational as well as application aspect of nonlocal correlations. We will consider the simplest setting for a Bell experiment which constitute two space-like separated parties, each performing one of the two possible measurements with binary outcomes. Then we discuss: (i) the geometry (boundary) of the set of quantum correlations, (ii) applications of these correlations in self-testing quantum devices, and (iii) distillation of weak quantum correlations to strong ones.