Physics Inverted Materials is commercializing machine learning methods to accelerate materials development. We are focusing first on accelerating atomic scale simulations to develop new materials that undergo reactions in the bulk and at interfaces. We are developing PHIN-atomic, a software tool that automatically trains a machine learning interatomic potential (MLIP) for running high accuracy molecular dynamics simulations. Our software is based on our innovative uncertainty quantification tools that improve the training of our machine learning models.