Dynamics of self-assembly and structural transitions in amphiphilic systems play an important role in various processes, ranging from production of nanostructured materials to transport in biological cells. Theoretical and computational modeling of these processes is extremely challenging due to the large span of length- and time-scales involved. In this talk, we discuss development of a multi-scale model for formation and disintegration of non-ionic and ionic spherical micelles. This study is performed under the assumption that the dominant mechanism of micelle formation (disintegration) is a stepwise addition (removal) of single monomers to (from) a surfactant aggregate. Different scales of these processes are investigated using a combination of coarse-grained molecular dynamics simulations, analytical and numerical solution of stochastic differential equations, and a numerical solution of kinetic equations. The removal of a surfactant from an aggregate is modeled by a Langevin equation for a single reaction coordinate, the distance between the centers of mass of the surfactant and the aggregate, with parameters obtained from a series of constrained molecular dynamics simulations. We demonstrate that the reverse process of addition of a surfactant molecule to a surfactant aggregate involves several additional degrees of freedom, including orientation of the surfactant molecule and micellar microstructure. Formation of ionic micelles involves, in addition, one more degree of freedom which describes collective dynamics of the charges in the system. Time-scales of the additional degrees of freedom are comparable with the time-scale of the monomer addition to a micelle and hence these degrees of freedom play an active role in the monomer addition process. We demonstrate that neglecting their contribution leads to qualitative discrepancies in predicted surfactant addition rates and propose a stochastic model for the monomer addition which takes the additional degrees of freedom into account. The model parameters are extracted from molecular dynamics simulations and the surfactant addition rates are determined from Brownian dynamics simulations of this model. The obtained addition and removal rates are then incorporated into the kinetic model of micelle formation and disintegration. It is expected that insights gained in the course of development of the multi-scale model for this relatively simple self-assembly process will aid in the development of models for dynamics of more complex amphiphilic systems, such as formation of cross-links between worm-like micelles in solution and fusion of lipid bilayers.