Difference between data scientist and machine learning engineer?

In the San Francisco / Bay Area, what is the difference between a machine learning engineer and a data scientist? More specifically, who is more likely to be responsible for prototyping and building machine learning models and performing feature engineering / analysis?

In the past and for tabular datasets (predicting bankruptcy, fraud, heart disease, marketing responses, etc), I would have said that the data scientist would be in charge of creating machine learning models and the machine learning engineer would be in charge of deployment. I’ve had a lot of interviews recently, and in San Francisco in particular it seems that most startups have taken the responsibility of predictive / ML models out of the data scientist role and given it entirely to the ML engineer. In many places it seems the data scientist role has become solely about A/B testing and analysis. Is this part of a broader trend or are my 10 or so personal anecdotes not representative of the industry at large?