As discussed here and here, the 1 GB memory restriction for the free tier of Cloud is not enough to install TensorFlow or compile Stan models. As a workaround, I would like to build all these binaries locally and ship them to Cloud over the network. To make sure the toolchains match up, I would like to borrow the Dockerfile of my Cloud project, build a local image, and compile what I need inside that local image. Is such a Dockerfile available?