As scientists' needs for computational techniques and tools grow, they cease to be supportable by software developed in isolation. In many cases, these needs are being met by communities of practice, where software is developed by domain scientists to reach pragmatic goals and satisfy distinct and enumerable scientific goals. yt is a visualization and analysis toolkit built around physically-motivated in- quiry of astrophysical data, deployed widely across XSEDE systems. Enzo is an adaptive mesh refinement platform for astrophysical simulations. We present techniques that have been successful in growing and engaging communities of practice, specifically in the yt and Enzo communities, and suggest that their successful application in the relatively small computational astrophysics community be viewed as a testbed for wider deployment in the computational science and XSEDE communities.