Optimal laser pre-sintering parameters are essential for reducing the surface roughness of laser melting samples. This paper investigates the impact of process parameters, including laser power, exposure time, line spacing, and point spacing, on surface roughness during the pre-sintering process. To achieve high-quality processing parameters, a response surface method (RSM) and a support vector machine (SVM) model optimized by the warp snake algorithm (SO) were established. The results indicate that both models possess strong predictive capabilities, with the SO-SVM model demonstrating superior optimization and generalization performance. The minimum surface roughness achieved through the SVM model optimized by the snake algorithm for the refined process parameters is 17.7 μm, which is lower than the 19.3 μm obtained using the response surface method. This study provides a reference for minimizing surface roughness, significantly reducing the trial-and-error costs in the machining process, and facilitating the production of higher quality machined parts.