SUMMARY
The existence of causal relationships between a material’s processing conditions, its microstructure, and its mechanical properties is a foundational tenet in the field of materials science and engineering. These relationships are usually investigated via data-driven process-structure-property (PSP) relationships which provide a modular and hierarchical framework for materials knowledge curation. Establishing high-quality P-S or S-P linkages for a specific material relies highly on human data-science expertise hindering its wide application in the materials domain. This becomes a bottleneck for the efforts to understand the material's genome for designing advanced materials. This work proposes an automated machine learning workflow/tool that will reduce the human effort for establishing S-P workflows. This autoML tool will improve the performance and the reproducibility of the P-S-P workflows. This work proposes three tasks to develop an autoML tool for P-S-P surrogate models: 1) Memory and time-efficient workflows for P-S-P surrogate models via “distributed” PyMKS, 2) Graph representation for P-S-P surrogate models (workflow) for reproducibility, 3) Auto S-P surrogate model generation workflow (AutoMKS).