Surface modification using specialty layers, such as carbide composites (WC-Cu, WC-Co) and DLC (Diamond-Like Carbon) coatings on structural steel substrates, is a key element in improving the service life of technical components. However, the laser micromachining process of these layers is characterized by high nonlinearity and significant sensitivity to uncontrolled factors, which makes precise control of their target microgeometry and tribological properties difficult. This paper presents a comprehensive approach to the mathematical modeling of the process, with controlled factors including laser power, operating mode (pulsed vs. continuous), and feed strategy. The observed variables were the geometric parameters of the microtexture and the coefficient of friction determined using the ball-on-disc method. The research methodology is based on classical design of experiments (DOE) in the RSM variant and factorial models, which were extended with advanced uncertainty quantification analysis. To reliably assess the model's predictions, three approaches were used in parallel: classical asymptotic formulas assuming a normal distribution of errors; nonparametric bootstrap resampling methods, which allow for the determination of confidence intervals without arbitrary statistical assumptions; and a priori uncertainty estimation based on fuzzy logic to describe the influence of uncontrolled variables. The research results indicate varying sensitivity of the tested materials to irradiation parameters, with DLC coatings demonstrating specific response characteristics in the pulsed operation range. It was shown that the use of nonparametric and fuzzy methods yields significantly more realistic error estimates than the classical approach, particularly in the boundary regions of the process parameters. The developed methodology is a tool for optimizing laser micromachining with enhanced technological reliability, which is crucial for precise surface engineering.