Is Tensor Flow Compilable to a binary?

If I install all the external dependencies of Tensor Flow such as the Bazel, make, on my linux machine, and build Tensor Flow from source, would I be able to arrive at a binary code I can feed into some server machine to just run?

I am also asking whether the python API calls a bunch of dynamically linked libraries, and then somehow the have a model.build() command or a model.compile() command that produces a binary in the directory in my linux machine?