This is a notoriously difficult topic to pin down. For people that genuinely want to tease it apart, there are useful ways to do it, such as: http://www.datas

This is a notoriously difficult topic to pin down. For people that genuinely want to tease it apart, there are useful ways to do it, such as: http://www.datasciencecentral.com/profiles/blogs/six-categories-of-data-scientists http://www.datasciencecentral.com/profiles/blogs/six-categor... A bit of transparency goes a long way. Where I work (Casetext), our data team does a mix of traditional data engineering, machine learning, graph algorithms, NLP (a mix of off-the-shelf and custom work), information retrieval, customer/market analytics, and experimental design. For us, and probably many other companies in this situation, it helps us to know what a candidate is interested in and good at. It also helps explain what we do to another level of granularity. Yes, the "data science" label is pretty muddled. It happens (but it is understandable) probably because of the fuzziness of the topics and differing communication styles and goals of engineers, marketers, salespeople, and investors.