How to apply operator level optimizations

I understood importing other framework model into TVM relay, and applying different graph level optimizations.

# Import model into relay
relayModule, relayParams = relay.frontend.from_tensorflow(graph_def)
# Apply differnt graph level optimizations
relayModule = relay.transform.SimplifyInference()(relayModule)

I also understood compiling whole relay module using ‘relay.build’ and running the model using graph runtime.

My question is how to apply different operator level optimizations individually using tensor expressions scheduling primitives for above relay module? Is there any Python API which will convert relay module into tensor expression module, to play with tensor expressions?