It doesn't matter much because all the real computation happens on the GPU. But you could take their neural network and do inference using any language you want.
