I was recently reading an interesting report by O’Reilly called The Care and Feeding of Data Scientists. This report discusses various topics, such as how to hire data scientists, how to organise data science teams and how to help data scientists grow.
Something which I found particularly interesting was a section talking about the different types of data scientists. This is a topic that is particularly close to me, as I have discussed it in the past, but I am also talking about it on my workshops and seminars.
O’Reilly provides a slightly different break down of the different types of data scientists, to the one I have discussed in my article The Tribes of Data Scientists. I believe both breakdowns to be valid, and complementary to each other. Data science is a new field, so you will get different perspective on what a data scientist should do, depending on whom you ask. So, let’s discuss O’Reilly’s categorisation of data scientists.
The different types of data scientists
O’Reilly’s period table consists of the following types of data scientists.
Operational data scientists
These are data scientists who understand the business and its systems inside-out. They translate technical problems into business goals, and help a business improve its efficiency or better meet its goals through data science. This role sounds a bit like a business analyst, but the main difference lies in that an operational data scientist is more likely to be doing more advanced modelling, whereas a business analyst might more focused on using simpler tools, like dashboards.
Product-focused data scientists
These are data scientists who are focused on developing or improving a product using data science. They might be even embedded within the product team of a company. A good example of this type of data scientist is someone who is developing a recommendation system for a company’s website.
Engineering data scientist
This is a more technical role. This type of data scientist closely resembles that of a data engineer (some people might argue they are even the same thing). Engineering data scientists are responsible for making sure the systems run smoothly and they create pipelines that can work at scale.
Research data scientist
This describes a data scientist who is very much focused on discovering new techniques and algorithms. The archetypal example of this data scientist is someone who has finished a PhD in Machine Learning and is working in a company like Google, Deep Mind or IBM. This is the most glamorous type of data scientist, since they have an excellent understanding of data science, and heavy academic credentials. However, in practice, most companies do not really need their full set of skills, and might not even have the infrastructure to support them. It is common to see companies hiring someone with this set of skills, where in reality all they need is an operational or product-focused data scientist.
What type of data scientist do you need?
Finding the right person to hire is not easy, especially when you are not sure what kind of skills you are looking for. Hopefully, this article helped you get a better understanding of what the landscape looks like in data science. The Tesseract Academy has also created a very useful tool for identifying the skills you need in a data scientist: The HDS Questionnaire. If you are planning to hire data scientists, you will find it extremely useful. Get in touch if you have any questions or comments.