Telescope simulations are an invaluable tool for astronomers and can be used to optimize design parameters and observing strategies of real instruments, forecast science goals, and develop and test data reduction algorithms. New millimeter and sub-millimeter observatories coming online in the next decades are characterized by a wide range of designs and huge data acquisition rates, and using them effectively will require a flexible yet comprehensive simulator that is as lightweight as possible but also capable of scaling to large-scale distributed computing in order to emulate the large throughput. We introduce a new simulation framework called \tt{maria} which comprises a full stack of all the components that affect raw time-ordered data collected by telescopes. \tt{maria} is capable of simulating both bolometer and interferometric observations at a wide variety of instruments and sites around the world. We provide an overview of each component of the simulation stack, focusing on a more computationally efficient model for simulating fluctuations in atmospheric emission. We also demonstrate the flexibility of \tt{maria} provide a number of examples of diverse simulation scenarios for CMB telescopes like ACT and SO, interferometers like ALMA and the VLA, and large single-dish telescopes like AtLAST and the GBT.