Contributing on GPs

At first, I wanted to contribute to edward adding a GP regression example (with the classic Mauna Loa CO2 dataset). But it seems there isn’t implementation of other kernel than RBF, would it be a contribution of interest to add new kernels to edward ?
Also regarding kernels it seems that hyperparameters aren’t learned, is it work in progress ? This is also an area where I would be happy to contribute.