Abstract: In this talk, I present the development of various models for dust physics suited for galaxy formation simulations.  I begin by introducing a model to evolve the spatial distribution of dust in galaxies, accounting for processes that affect the interstellar dust budget, like stellar dust production, accretion of gas-phase metals, and supernova-driven destruction.  Using the moving-mesh hydrodynamics code AREPO, I perform cosmological zoom-in simulations of Milky Way-sized galaxies to study the evolution of interstellar dust.  Predictions from this model compare favorably to a number of observed low-redshift dust scaling relations.  I also present simulations of uniformly sampled cosmological volumes to analyze the behavior of dust statistics on large scales.  Next, I develop a model to more realistically track the dynamics and sizes of interstellar grains.  This novel framework handles dust using live simulation particles, each representing a population of dust grains of different sizes and subject to dynamical forces like aerodynamic drag.  Using simulations of idealized galaxies, I illustrate how different physical processes affecting dust grain sizes would impact galactic extinction curves.  Finally, I describe an extension of these methods to couple dust physics and radiation hydrodynamics in the code AREPO-RT.