A primary objective of the SciDAC project "Building a Universal Nuclear Energy Density Functional" is to develop a microscopically-based, quasi-local energy density functional (EDF) in the form of a generalized Skyrme interaction with theoretical error bars for fine-tuning to data. In this talk, I describe one possible approach for accomplishing this objective based on applying the Density Matrix Expansion (DME) of Negele and Vautherin to Renormalization Group (RG)-improved two- and three-nucleon interactions from chiral effective field theory.