Embodiments of the invention provide a system for performing dynamic execution of data processes in distributed server systems. The system is configured for identifying initiation of at least one data process that is scheduled to be executed on at least one server of distributed server systems associated with an entity, communicating with a metrics node comprising a metrics machine learning model to gather information associated with the at least one server, determining that the at least one server will experience degradation of performance while performing at least one step of the at least one data process based on communicating with the metrics node, and routing the at least one step of the at least one data process to at least one new server and execute the at least one step on the at least one new server.