Embedded, Class 0/I protostellar disks represent the initial condition for planet formation. This calls for better understandings of their bulk properties and the dust grains within them. In this poster I summarize a few recent results on these topics. First, we perform global radiative non-ideal MHD simulations covering the collapse of the pre-stellar core and the first ~10 kyr of protostellar disk evolution. These simulations suggest that typical Class 0/I disks are gravitationally self-regulated, where gravitational instability triggered by the envelope infall act as the main heating and angular momentum transport mechanism and most of the disk is marginally unstable (Toomre Q = 1~2). This leads to a simple disk model where disk properties are constrained by marginal gravitational instability and thermal equilibrium. We then show that observations of Class 0/I disk populations are consistent with this model, and fitting our model to observation suggest disk-to-star mass ratio of order unity. We further propose simple methods to empirically test whether a given disk is likely gravitationally self-regulated and to constrain grain growth in gravitationally self-regulated disks. Finally, using observational constrains of typical grain size and semi-analytic grain coagulation calculation, we argue that the earliest stages of planet formation probably involves a bimodal grain size distribution where most grains are capped below the fragmentation barrier but a small fraction of grains grow into much bigger sizes and eventually form planetesimals and planets.