Language Model on Code dataset

I was wondering one of the main challenge that I face while doing software development is assigning name to the variables. What are your thoughts on applying a language model on the dataset of well written code (can be obtained from github i suppose) and provide suggestions to the variable name as and when typing. What could be flaw in this thinking and things to consider before starting something like this?