We consider solving multiobjective simulation based optimization problems in renewable energy and imaging type settings. Predominantly motivated by machine learning, we consider both classification and regression type regimes. Our objectives of interest range from accuracy and computational time to algorithmic bias and sparsity. Models can be comprehensive ranging from complex decision trees and neural networks to simple KNN type settings. Multiobjective derivative-free optimization has deployed Bayesian and Direct-Search methods individually in the past. In this work, we propose a switching framework that effectively uses both these methods in an iterative fashion. Secondly, we specifically analyze machine learning type problems and propose a warm-start based training methodology that effectively deploys the problem structure. We test and compare the performance of our methods on real world problems pertaining to energy markets and imaging. We observe a significant promise in numerical performance of our joint switching based scheme.