This paper presents an approach for the optimization of continuous manufacturing process parameters, as well as the creation and deployment of predictive models for rapid adjustment of production assets by operators. The approach developed was implemented based on the integration/automation of simulation processes, design optimization and predictive modelling (ML/AI) in a collaborative cloud environment. We illustrate the effectiveness of the approach on a real problem of pultrusion of glass fiber reinforced C-sections. The optimization parameters are the initial resin temperature, the temperature of the first die zone, the temperature of the second die zone and the pull rate. The constraints are the transverse stress in the pultruded profile, the maximum material temperature, and the degree of cure at the final section. A bi-objective optimization problem is considered: the pull rate and the degree of cure must be maximized. Multi-objective optimization based on metamodels was used. A second task is to build a (fast) approximation model using a machine learning technique and deploy it via web services. Operators setting up production machines can then use the deployed model to predict production quality and set up their machines to increase productivity. Building the model and verifying its quality/reliability is a matter for experts, but once the model is validated, it can be used by any discipline engineer or operator who directly adjusts the production equipment. The challenge then is to offer such a model in the form of an easy-to-use calculator and then deploy it under the supervision of the expert who may want to update the models characteristics from time to time. The link with the expert must not be broken. The solution to this challenge is to use a collaborative environment where the model owner (the expert) and the operator can access the same model. A Cloud architecture was used to enable this deployment and collaboration. The model is built in the Cloud and presented as an automatically generated web application that provides the necessary simplification layer for operators. Operators can then evaluate any combination of their machine parameters and get an instant prediction of production quality.