The difference between a scientist and engineer is prototyping vs productionizing
Data scientist - use statistical and ml techniques to build models on top of data
Data engineer - use software and processing tools to build pipelines for data
ML scientist - use state-of-the-art ML models to prototype ML models
ML engineer - use high-level ML framework to train and productionize ML models
There are roughly twice as many open engineering roles vs scientist roles
With frameworks like Tensorflow and PyTorch, setting up the prototype is getting easier but the "boring" skills like building an ETL pipeline is getting increasingly scarce