Dark matter makes up most of the mass of the Universe, yet it remains elusive. In this talk, I show how we can image the distribution of dark matter on small and large scales using stellar/gas kinematics and strong gravitational lensing. Confronting these data with numerical simulations gives new constraints on the nature of the dark matter fluid. However, to provide competitive constraints, two key challenges must be overcome. On the data modelling side, I show that it is vital to properly marginalise over model degeneracies. I stress the importance of running end-to-end tests of our methodology on dynamically realistic mock data. On the theory side, I discuss the importance of correctly modelling "baryonic processes" like stellar feedback. Such feedback could significantly reshape the dark matter distribution inside galaxies, leading us to false conclusions about the nature of dark matter. I conclude with a look to the future and what we can hope to learn with the advent of Euclid, the LSST, and Gaia.