When looking to hire or contract with a data scientist or consultant, it’s important to understand the types of skill sets that “data scientists” typically have. In talking with colleagues across the industry, I’ve found that they typically fall into two categories.
The first group of data scientists are hired to design, build, and optimize “learning” systems. They have a penchant for deep, rigorous thinking and a taste for solving difficult problems. I call this group “algorithm scientists”: they thrive under the challenge of understanding the minutiae of the systems that they build, and often have a master’s understanding of math and computer science.
A good (and brilliant) friend of mine from grad school falls into this category. As the lead data scientist for a streaming music company, he built and refined their music recommendation engine. The algorithm scientist should be challenged with hard problems with longer horizons — for example, improving Amazon’s recommendations by 0.1% would potentially drive millions in new revenue.
The second group of data scientists go broad — they tend to be generalists, working across an organization to solve business problems. I think of this group as “business scientists”: they apply scientific reasoning to test and optimize a company’s business model and operations. While the Algorithm Scientist sits at the confluence of Math and Computer Science, the Business Scientist has a better understanding of how to dissect business problems.
The person who excels in this job is fundamentally different in skill-set and character than the Algorithm Scientist: this person is more of a “hacker”, doesn’t form deep connections to a problem space, and likely has a strong entrepreneurial streak. This person should be challenged to work across silos, connecting disparate parts of the business.
I definitely consider myself a Business Scientist. In graduate school, I spent many years building very beautiful, but ultimately untestable mathematical models of the universe. When I took my first job as a data scientist, I realized that the business faced difficult decisions every day, and those decisions were often made by the highest ranking person’s instinct. Adapting to my new career, I immediately learned how much I loved changing that person’s opinion with numbers.
Ultimately, your hiring plans should include an honest assessment of the types of problems you experience in your business. This will make your engagement with a data scientist more productive for both of you!
Esoterica 