摘要： Laser beam welding is a non-traditional, advanced technique used for similar & dissimilar materials and is widely used in various industries like automobile, aerospace, nuclear reactors, etc. at a faster pace. As it is a complex process, it is very difficult to find the optimal process parameters. The primary point of welding is to acquire a high quality joints and requiring little to no effort. However, without optimization, it is impractical to accomplish minimal effort welding. The principle work of this exploration is to display, break down and improve weld bead geometry in the powerful ND: YAG laser butt welding of Inconel 625. However, in view of the literature review, the responses are continuous and have impact on the welding geometry. In this way the feasible varieties of the judgment factors are started by coordinating trial tests. The Design of Experiment (DOE) system is used to generating the experimental plan and then conducting the experiments according to the plan. After recording the responses Reaction Surface Methodology (RSM) is received for precise expectation numerical models to evaluate the response variables and are created from the experimental information. Artificial Neural Networks (ANNs) have adopted from the MATLAB software for analysing the output response regression and provide the best curve fitting among the input and output variables.