I propose to develop a broadly applicable, quantitatively reliable, computationally simple approach to the study of large heterogeneous systems, and to apply it to important problems in molecular and organic electronics and photovoltaics. This will be based on a radically different approach to the development and application of density functional theory (DFT) - determining an optimally yet non-empirically tuned system-specific functional, instead of seeking a universally applicable one.Large heterogeneous systems are vital to several of the most burning challenges facing materials science. Perhaps most notably, this includes materials systems relevant for basic energy sciences, e.g., for photovoltaics or photocatalysis, but also includes, e.g., organic/inorganic interfaces that are crucial for molecular, organic, and hybrid organic/inorganic (opto)electronic systems. Theory and modelling of such systems face many challenges and would benefit greatly from accurate first principles calculations. However, the “work-horse” of such large-scale calculations – DFT – faces multiple, serious challenges when applied to such systems. This includes treating systems with components of different chemical nature, predicting energy level alignment, predicting charge transfer, handling weak interactions, and more. Solving all these problems within conventional DFT is extremely difficult, and even if at all possible the result will likely be too computationally complex for many applications.Instead, I propose a completely different strategy - sacrifice the quest for an all-purpose functional and focus on per-system physical criteria that can fix system-specific parameters without recourse to empiricism. The additional flexibility would help us gain tremendously in simplicity and applicability without loss of predictive power. I propose a practical scheme based on tunable range-separated hybrid functionals and a plan for its application to a wide range of practical systems.