How to Find — and Keep — the World’s Best Data Scientists

In a world where geek is chic, data scientists — particularly those specializing in cutting-edge machine learning — are downright sexy. This small group of individuals has become an elite class of professionals being courted by a growing number of companies.

According to McKinsey‘s May 2011 report, the U.S. has a shortage of 140,000 to 190,000 people with deep analytical skills. So companies must find new ways to acquire them and keep them happy. Below, we discuss what we’ve learned as we’ve recruited — and retained — more than 230 leading machine-learning scientists.

Creating a Critical Mass of Scientists

Scientists prefer to work in a community of like-minded individuals with whom they can collaborate, learn from and feel comfortable with. In addition, they want to know that they’re a valuable and important group within the organization. And companies that already have a critical mass of scientists find it easier to attract more.

Several things have worked for us in this regard. First, we created a scientific advisory board that includes global leaders in machine learning, mostly from academia. This is a working board; we go to them with our hardest problems and bring them together several times a year to hear what they’re working on. The advisory board gives us access to talent coming out of universities around the world, and our association with intellectual leaders is appealing to potential recruits. Well-publicized contests are another excellent way to acquire new talent. By entering (and winning) competitions, your company can dramatically raise its profile, helping good scientists find their way to you. Finally, you can use recruiters, but be sure to use only those who truly understand your business.

Once you’ve selected potential candidates, senior-level scientists should be highly involved in the interview process. They need to put the candidates through rigorous interviews in which candidates are asked to explain their work and prove that it’s original. We often make prospects defend their theses.
To Know Them is to Keep Them

Being a data scientist in a business setting requires multiple skill sets. Success depends not only on one’s talent in applying analytic techniques, but also on one’s ability to understand the business problem they’re solving. To do this, they need a certain level of knowledge on the subject, which may include not-so-scientific areas, such as fashion, the auto industry or politics.

They also need a certain ability to understand human relationships, social networking and customer sentiment. In other words, the most successful scientists are well-rounded, curious and as engaged with the real world as they are with the abstract one.
Scientists Love to Learn and Hate to be Bored

This is probably not a huge surprise, but you need to foster an environment that encourages continual learning. Granting scientists easy access to conferences, offering incentives for self-improvement and published research articles, and providing challenging training programs can all be effective. But even these steps aren’t enough: New challenges must also be integrated organically into the work.

Scientists are problem solvers by nature, so they need new and interesting problems to attack on a regular basis. Their work should be varied and require multiple areas of expertise. Different projects — and different types of projects — will keep scientists engaged and challenged.
Data Scientists are Social

Data scientists need a sense of community when they work. A collegiate environment where they can explore solutions together helps keep them engaged and sharp. To foster community, we often host gatherings and work-centered presentations that encourage employees to meet, converse and learn about one another.
Data Scientists Need to Matter

Locking scientists in a back office and occasionally throwing them problems to solve is a recipe for low hiring rates and high turnover. They need to make a difference and be central to a company’s success. They’ll be most happy in a business where the science is the business and their work has a direct impact on the company’s bottom line and overall goals. For example, all our business units are co-headed by a general manager and a scientist. Science is that important to our success.

In this same vein, data scientists need a clear career path. If the company puts science at the forefront of the company, with equal or greater importance to the business practices, the scientists will get the appreciation they need. You can do this by structuring your organization so that the scientists have a track that parallels that of business professionals  – so that they can also end up at the top.

Data Scientists Need the Time and Tools to Explore their Curiosity

Micromanagement will not go over well with data scientists. Curiosity is a big part of the job, and they need the freedom to satisfy their own natural curiosity as well as the technology that takes the friction and drudgery out of their work. By not defining their roles too narrowly and by giving them the tools that allow them to get the most from data, they’ll uncover ways to better the business – even if that means solving problems you didn’t know you had.

By understanding all the dimensions and dynamics of data scientists, you’ll have a better shot of finding and retaining them. In a world that increasingly revolves around big data and the value that lies therein, companies able to retain the data scientists that hold the key to unlocking this value will find the most success.

Jacob Spoelstra is the global head of R&D at Opera Solutions. With more than 19 years of experience in machine learning, he focuses in particular on neural networks. Prior to joining Opera Solutions, he led a custom fraud analytics consulting team at Fair Isaac and held analytics leadership positions at SAS, ID Analytics and boutique consulting company BasePoint. Jacob holds B.S. and M.S. degrees in Electrical Engineering from the University of Pretoria and a Ph.D in Computer Science from the University of Southern California. You can reach him at jspoelstra@operasolutions.com.

Joseph Milana is the global head of Analytics at Opera Solutions. His areas of expertise include statistical modeling, data mining, online marketing, and fraud detection. Before joining Opera Solutions, he was a chief scientist at Fair Isaac, working in research and development and leading new product initiatives and core algorithmic development. He holds a Ph.D in Theoretical Physics from SUNY at Stony Brook and a B.S. in Applied & Engineering Physics from Cornell University. You can reach him at jmilana@operasolutions.com.

Sarah Anderson is a senior editor at Opera Solutions. With more than 12 years of editorial experience in tech publishing, she focuses on leveraging big data and predictive analytics to improve business growth. She has held editorial positions at PC Magazine, Laptop Magazine and Computer Shopper. Sarah holds B.S. degrees in English and Business Management from the University of South Dakota. You can reach her at ûsanderson@operasolutions.com or follow her on Twitter at @seander01