Successful fabrication of intermetallic coatings on surfaces of manufacturing interest involves regulation of the temperature/concentration dynamic distributions that develop in the molten layer during the thermal and reaction process. Modeling the spatio-temporal dynamics of this metallurgical process, however, requires partial differential equations that are cumbersome to solve on-line, as part of a real time reference model to the controller. To this end, we present a computationally parallel and meshless model (i.e., decoupled with the capability to be solved numerically in real time) to decipher the dynamics of the thermal coating process and to permit real time monitoring and control of the resulting coating microstructure. The analytical model is based on kinetic growth theories, lumped energy and mass balances, and convolution expressions of distributed temperature and concentration Green’s fields (accounting for the orientation of their gradient and decomposing heat and mass transfer across the coating from substrate conduction). The model is validated with nickel aluminide coatings processed on a robotic plasma arc laboratory station, through in-process infrared thermal sensing and off-line metallographic analysis. A Monte Carlo sample control scheme, that involves on-line parameter identification and model adaptation, is also developed using the model as an in-process observer for successful production of binary metal system coatings that exhibit the desired microstructure geometry and characteristics.