The Data Scientist Shortage is Huge. Here’s How to Beat It.

In this special guest feature, Roberto Reif, Executive Director, Professional Development at Metis, discusses how the worldwide deficit of data scientists is real, but now that you’ve started thinking about your strategy and how you’ll resource the best people to help you execute it, it won’t seem so intimidating. Metis is a company that provides full-time immersive bootcamps, evening part-time professional development courses, online learning, and corporate programs to accelerate the careers of data scientists. Roberto came to Metis from Sensoria Inc., where he led the signal processing team, and has worked in applications in the healthcare, internet of things, and business intelligence markets. Roberto received a PhD in Biomedical Engineering from Boston University, and has co-authored several scientific publications, book chapters, and patents.

It’s no secret that employers are looking for data scientists. Businesses of all sizes have woken up to the fact that data science has the potential to drive efficiencies, mine new insights from decades of accumulated data sets, and otherwise transform their businesses. From Zillow’s home price predictions to Amazon’s recommendation engines, applications of data science have become increasingly prevalent and high-profile.

But while data scientist has been ranked the “#1 Job in America” for three years running, according to careers website Glassdoor, there’s still a shortage of talent to fill the huge need employers have. In fact, according to a recent LinkedIn study, businesses across the nation need 151,717 more data scientists right now.

Faced with this dire shortage of talent, business leaders who want to make the most of data science can’t rely on half measures and casual hiring processes. What they need is (1) a strategic roadmap toward building data science skills and (2) an effective hiring and resourcing plan.

In coming up with a roadmap to use data science effectively at your company, you need to first examine your specific needs. The fact is that many businesses, especially small- and medium-sized businesses, don’t actually require cutting-edge data science techniques to get a leg up on the competition. Still, there are some skills that every data science function at every company needs. You can think about data science team skills as on a continuum spread from the business side to the engineering side, as seen in the image below:

Are there areas on this spectrum where your team is particularly weak? Identifying where your resources are lacking and where you’re already strong is key to building out your roadmap to taking advantage of everything data science has to offer.

Once you’ve figured out your skill-building strategy, your next step is to hire for and resource the roles that will support it. As the data science space is relatively new in comparison to other fields in the business world, you may have some trouble sorting out all of the different titles created in recent years to identify what different data science roles do. Don’t get too hung up on titles; in some companies, one person fulfills many roles. Here’s a look at some of the main titles you’ll encounter when building out your data science function, arranged along the same continuum of business to technical roles as the skill sets were in the image above:

Hiring for the roles that match up with the skill areas you’re currently lacking will help you head down your strategic roadmap. Just remember that in the hiring process, finding the right person does not come down to who can solve every tough data problem you throw at them in an interview. It’s much more useful to be highly selective about figuring out which problems can impact your business most, then hiring relevant candidates to support solving them.

One major mistake that businesses make when starting a data science team is ignoring their strategic roadmap in favor of pursuing a “unicorn” candidate. Popular terms under the categories Big Data (like Spark/Hadoop), Deep Learning/AI (including Keras/TensorFlow), and Machine Learning (like Reinforcement Learning, Natural Language Processing) are in vogue as they’re relatively new techniques and platforms. And it’s not uncommon for executives to hear the buzz, then decide that they must need candidates focused on the latest-and-greatest too. But job descriptions that include all or most of the terms above rule out candidates you actually need to solve your specific business problems (plus you’ll have to pay $250k+ for the rarity of their skills).

The key is to be laser-focused on your business problems at hand. Are you looking to optimize your inventory and distribution forecasting ability? If so, a person with a solid understanding of statistics and experience building regression models using time-series data may be all you need. Remember that you don’t need someone with a PhD in Stats or Computer Science with 10+ years experience to do this; there are many entry-to-mid-level candidates from diverse STEM backgrounds who possess a strong working knowledge in these areas.

If you’re still unsure where to begin and how to diagnose your data problem, try and find someone who successfully led a data team that has done something similar to what you’re looking to achieve. They can help you understand your problem, how it translates into the data you need to collect and analyze, and what types of various data science skills you’ll need to hone in on when searching for candidates. This type of early-stage advice can help you avoid many of the pitfalls that businesses face when looking to hire talent for their new data science team.

The worldwide deficit of data scientists is real, but now that you’ve started thinking about your strategy and how you’ll resource the best people to help you execute it, it won’t seem so intimidating. Keep your eyes on the prize — people who can solve your specific problems — and you’ll be able to win the fight against the data scientist shortage.

Sign up for the free insideBIGDATA newsletter.