I am looking for a program that I can deploy on my computer or even better on my own cluster (Amazon EC2, OpenStack, etc.) that would offer Machine Learning as a Service (MLaaS). Ideally the program would be free and open-source.
To put it more simply, the program would take some dataset as input and try a bunch of different machine learning algorithms (e.g. NB, KNN, ANN, SVM, DT, etc.) with different parameters (number of clusters/neurons/layers, activation function, metrics used, regularization, etc), then output the best models it found. (best being defined using some usual metrics like accuracy, AUC of ROC, F1, etc.).
I know that it's easy to test a bunch of different models with some existing libraries like scikit-learn, but I'm looking for a solution that requires no or super-minimal coding and that would be as parallelizable (multi-CPU/GPU) and distributable (deployable on a computer cluster) as possible.