Complex phenomena, including superconductivity, magnetism, phase separation, and competing phases, emerge as many correlated atoms are collected together. Conventional theory has made limited progress towards a systematic and reliable way to study even the simplest models of these systems. We us massively parallel Quantum Monte Carlo methods, along with effective medium theories, to study these model systems with some success especially for cuprate models. The main limitation of these methods is minus sign problem which makes the study of strongly correlated system NP hard. Supercomputer power alone cannot be used to overcome these problems. To circumvent this limitation, we are developing multi-scale methods which separate the problem into different length scales, each with an appropriate approximation. Strong correlations at short length scales are treated explicitly, weaker correlations at intermediate length scales are treated with diagrammatic approximations such as the parquet, and the weakest correlations at long length scales are treated in a dynamical mean field. The parquet approximation, used for intermediate length correlations, scales algebraically with system complexity and scales efficiently to thousands of processors. Together, this multi-scale many body approach holds the promise of new discovery in and understand of strongly correlated systems.