Method, system and computer program product for optimizing the deployment of analytical pipelines

A computer-implemented method for optimizing the deployment of an analytical pipeline in an industrial production environment, is provided. The method comprises: providing an analytical pipeline (1) to be deployed and a description of an available infrastructure (2) for the deployment of the analytical pipeline (1), wherein the analytical pipeline (1) comprises a plurality of analytical phases, wherein the description of an available infrastructure (2) comprises a plurality of computing elements located in a cloud layer and a plurality of computing devices (80₁-80_N) located in an edge layer; at an optimization module (5) implemented in a computing device: from the analytical pipeline (1) and the available infrastructure (2), extracting a set of variables x = [X_f1, X_f2, .., X_fM], and providing the set of variables to a multicriteria solver (503); at the multicriteria solver (503) executing a machine-learning algorithm configured to, using the provided set of input variables x = [X_f1, X_f2, .., X_fM], provide an optimal deployment for the analytical pipeline, wherein the machine-learning algorithm minimizes a cost fitness estimation function and maximizes a performance fitness estimation function and a resilience fitness estimation function, thus minimizing the cost of the deployment while maximizing its performance and resilience.