You can use the SQLExecutor to create the indexes, e.g.:

CREATE INDEX index_name ON table_name (column_name)

If you want to make sure the SQLExecutor is executed last, use e.g. a FeatureHolder and then a Sampler (1st feature) before the SQLExecutor.

You could also put the SQLExecutor into a child workspace and have it called with another WorkspaceRunner from the master workspace.

@honkovam I second what david_r suggested to create the index like eg

CREATE INDEX index_name ON table_name (column_name)

but I usually put it into the navigator panel in the SQL Statement to execute after translation parameter if you wanted to cut down on having workspace runners

Or just use the FeatureWriter, that is if you are using FME 2016, and after it the SQLExecutor