ML tends to be more concerned with scale and computational efficiency as the algorithms used are often (a) heavyweight and (b) less human interpretable. Success is primarily measured objectively but via proxy measures like prediction accuracy. It may require data science efforts behind the scenes in order to succeed.
Data Science tends to be more concerned with interpretation than scale or repetition. It sits somewhere between analyst, sataistician, and journalist. It's also a much more interactive workflow often involving a great deal or successive, simple models and lots of exploratory visualization. Measures of success include discovery of human-relatable inferences from data, substantiation or dismissal of human-relevant experimental hypotheses, and success in communication of those ideas.
Statistics is a field of study of some of the most important mathematical foundations of each of these roles. A statisicians skill is more generalized but also more abstract. A statistician who learns CS and journalism could easily head into either other field.
And what's the difference between data science and statistics? Statisticians try to come up with new mathematics, while data scientists just try to come up with ways to apply that maths?