Used in industries such as automobiles and aeronautics, metal laser welding remains a challenging process to master. It is characterised by its high penetration/width ratio, useful when the weld is realised through a material. As the result of a physical transformation involving non-linear laser-matter interaction, thermodynamic and fluid mechanic, it possesses numerous interactions between its parameters, making it a poor candidate for simulation or Design of Experiment modelling. The main method to get the production-ready set of parameters remains the time-consuming and labour-intensive trial and error. In this study, Artificial Intelligence is investigated on readily available data from previous trials to predict weld penetration. We found that feature engineering, and especially data augmentations, significantly improves the prediction. Tested in challenging situations, the model highlighted its abilities both in interpolation and extrapolation on most of the unstudied materials. Such a model could be used to identify the set of parameters able to achieve a given laser weld penetration in new metals.