Metal additive manufacturing (AM) holds transformative potential across many industries. Yet, identifying optimal processing parameters remains a key bottleneck hindering broader industrial adoption. Traditional approaches often rely on trial-and-error experimentation, guided by expert intuition and basic statistical tools—methods that are time-consuming, resource-intensive, and difficult to scale.

In this talk, we introduce a Bayesian machine learning framework for efficient optimization of process parameters in metal AM, with a focus on Laser Powder Bed Fusion. Leveraging Bayesian Optimization (BO), our method can identify near-optimal processing conditions starting from minimal experimental data. The approach is extended to multi-objective optimization, where we apply batch BO techniques to simultaneously optimize multiple conflicting targets. Our results demonstrate that this data-efficient, probabilistic strategy not only accelerates process development but also reduces experimental cost and variability. The proposed framework is general, scalable, and readily applicable to a wide range of materials, AM technologies, and optimization objectives—offering a path toward autonomous, intelligent control in metal AM.