In the context of the industry of small and middle-scale manufacturing, the process of optimized scheduling decision-making is not yet widely spread despite the large gains that may be achieved through a better scheduling of operations inside the workshop. We focus on batch production systems where the problem is modeled as a Job Shop and both the machines and the operators are considered as flexible resources. Resources flexibility and multiple resources requirements are the two main constraints we can find in an industrial context. However, resource availability and precedence constraints can be also considered. We are working on a problem with multiple objectives, such as lateness as one of the main objectives but also earliness, waiting time between operations, or the number of jobs in process at any time. We decompose the problem into three sub-problems: resources assignment, operation sequencing and finally operation timing. The approach consists in a cyclic two-stage variable neighborhood descent around the first two sub-problems. We additionally consider a data-driven approach in order to improve and better guide the VNDs.