Recent advancements in astronomical observations allow us to detect populations of disrupted dwarf galaxies, so-called dwarf galaxy stellar streams, around the Milky Way, and external galaxies. At the same time, cosmological baryonic simulations now have enough particle resolution to resolve dwarf galaxy streams down to stellar mass of $M_\star\sim5\times10^5\Msun$. In combination, we can start to explore, analyze, and model dwarf galaxy streams that have been evolved in the full cosmological environments for the first time. This thesis paves the way for this exciting new area of research in preparation for future surveys such as LSST and Roman Space Telescope. Using state-of-the-art FIRE simulations, we identify and compile populations of simulated dwarf galaxy stellar streams. With this set of simulated stellar streams, we analyze and model their progenitors from orbits and kinematics to the phase-mixing process. This set of simulated streams can also be used to study near-field cosmology, dark matter, and the formation history of the Milky Way. As an extension of the famous ``missing satellite'' problem, we can now compare populations of disrupted satellites in simulations and observations. We generate mock Dark Energy Survey and LSST observations of the simulated streams and determine their detectability. Finally, we use one of the simulated streams to test the method of constraining the tilt of the dark matter halo using the action-angle-frequency formalism. We then apply the method to the Sagittarius stream's leading arm and report the constraint on the shape and minor axis location of the Milky Way's dark matter halo.