Industrializing a tool based on Machine or Deep Learning presents a number of risks, including : extremely volatile technologies (Tensorflow/Pytorch, CUDA/CUDnn, etc), lack of reliable tools, difficulties in packaging the final tool into an automatable element within an information system, tracking and monitoring, dynamic deployment on GPU cloud providers, protection of personal data… Datalchemy offers its experience to support its customers and, starting from a Proof of Concept, to implement the industrialization work in accordance with the rules of your company’s systems management.