Hi,
I would like to work in this problem.
I am currently working towards a PhD in Machine Learning and have experience in problems of this type.
Still, it would be useful to discuss your issue a little more. Maybe there are some alternative approach that might fit you.
For instance, instead of using ensembles, it is common to use dropout in NN. It is a newer/faster alternative to ensembles of NN. In addition, given that you are trying to predict values over time, it s likely that a recurrent neural network might be better.
As I said, if you could send me a little more information on the kind of output you are looking for, I can provide you with a good solution.
Let me know what you think or if you have any questions.
I may adjust the price/time based on our discussion.