8 Tips for Hiring a Great Data Scientist

If you’re trying to fill one or more openings for data science pros, you know how hard that task can be. The average time to fill data science positions reached 62 days in April, with the hiring time for a senior data scientist taking 70.5 days on average.

Don’t expect things to get better anytime soon. IBM predicts that demand for data scientists, data developer,s and data engineers will reach nearly 700,000 openings by 2020.

Fortunately, there are a few things you can do to gain an advantage over others in the market. Here’s how to improve your hiring process and ensure that your next data science hire is truly great.

1. Define the Data Problems That Need Solving

To develop an accurate hiring profile in a competitive field such as data science, you need to have a clear idea of the problems that need to be solved.

For instance, there’s often a lot of confusion between the role of decision scientist and machine learning engineer, noted Jacqueline Nolis, principal data scientist at Nolis, LLC. But they typically fulfill different tasks and purposes.

“You don’t need someone with a PhD to draw insights from business data and recommend a location for a new factory,” she explained. Indeed, over the course of her 15-year career, she’s only used her advanced academic skills a “handful of times.” A professional with a bachelor’s degree, an understanding of statistics, business acumen and an analytical mind could solve those types of problems.

While you frequently need a true data scientist who’s comfortable with the underlying theories to help deliver analytics, you may be able to hire a Python programmer who is familiar with data science toolkits, depending on the business results you want to achieve, noted Ben Taub, CEO of data science and data engineering recruiting firm DataSpace.

2. Consider Your Company’s Maturity Level

Another thing to consider: Your company’s data science maturity level, since you’ll need someone with leadership and visionary skills to create a strategic roadmap and build a practice from the ground-up.

In fact, having no infrastructure in place to manage and support a successful data science initiative is the number one reason why data scientists are motivated to change jobs.

3. Expand the Talent Pool

A recent study by Burtch Works reveals why it’s so hard for companies in smaller markets and traditional industries to compete against the likes of Google, Facebook and Apple for top data science talent.

Currently, some 44 percent of data scientists are employed in the tech industry, which also pays the highest median salaries. What’s more, some 40 percent are employed on the West Coast, while 26 percent reside in the Northeast.

If the demographics are working against you, you need to be flexible about job descriptions, experience and offering remote work and telecommuting to expand the talent pool. Also, keep in mind that 45 percent of data scientists have five years (if not less) of experience, and that the field is mostly attracting young professionals.

“In some markets and industries there are so few candidates available, that employers need to hire for attitude and basic competency and train for everything else,” explained Brian Shepherd, executive recruiter for Burch Works.

“Look for someone who knows how to break down problems and has taken data science courses on their own to expand the pool of potential candidates,” Nolis advised. “For instance, if you need to hire generalist data scientists for your team, look for engineers from the petroleum industry or other engineering disciplines.”

4. Communicate a Vision

Even companies in traditional industries can attract top data science talent… if they create and communicate a compelling vision in job descriptions, outreach emails and encounters with passive candidates.

“Data scientists want to do ‘cool work,’ not run reports,” Nolis insisted.

What’s “cool” right now? Hot candidates want to solve interesting problems or have the opportunity to reinvent or disrupt an entire industry. Dangling career development, opportunities for professional growth, and the ability to use the latest tools for data analysis will also help you attract the best data science talent.

5. Screen Prospects In, Not Out

Forget keyword searches and screening checklists. “Old school” techniques that are designed to eliminate candidates will work against you when you need to identify an up-and-comer who can grow into a more advanced data science position.

Taub explores a candidate’s basic understanding of data science by asking questions focused on theory and problems solving approaches. If that goes well, he digs deeper to see how a candidate has actually applied a theory or modeling technique to solve business problems or create a strategic advantage. For instance, if a candidate is familiar with Random Forest, he might ask how they’ve used the algorithm to build decision tree models.

Asking “questions of substance” will not only determine if someone is the right fit for a specific role; they can help you quickly identify promising candidates who have the capacity to up-skill on-the-job.

6. Use Case Studies for In-Depth Evaluation

Giving candidates a problem to analyze over a weekend lets them demonstrate their ability to manipulate data, extract information, draw conclusions and explain their recommendations to business leaders. For example, the problem could require candidates to utilize analytics tools or write solutions using Python.

“I’m interested in seeing how they communicate when asking follow-up questions via email or talking through their analysis,” Nolis said. “After all, I can always train someone to use a different model. I want to see how they engage and solve real business problems.”

7. Move Quickly

Time is of the essence when the best candidates are receiving multiple offers and off the market within days.

“Companies that are successful in landing top talent move from application to offer within two weeks,” Shepherd noted. “If the phone screen goes well, the face-to-face interview takes place within 24 hours, followed by an offer.”

They also realize that requiring too many assessments or interviews can drive away passive candidates. To make the hiring process faster and more efficient, competitive companies administer tests that take about two hours (instead of 20 hours) to complete. Don’t want to miss out? Speed up your hiring process.

8. Be Willing to Pay Top Dollar

To put it simply, there’s no point in recruiting data science talent unless you are willing to offer competitive pay and benefits. According to Dice’s data, the average data scientist makes $95,404 per year, and that’s before you plug in perks and benefits (or consider that some industries, such as finance or defense, pay a hefty premium for all talent.)

Nor is Dice alone in pegging data scientists’ salaries as notably high. According to the Burch Works study, the mean salary for individual contributors ranged from $94,987 to $168,170 in 2018, depending on experience level, while the mean salary for data science managers ranged from $146,133 to $247,633. Meanwhile, Glassdoor estimated the median base salary for entry-level data scientists at $95,000, making it the highest-paying entry-level job in the U.S. (in that company’s estimation).

And predictive analytics professionals command top dollar, too. “You’ve got to be in the ballpark on salary,” Taub warned. “Someone who is making $250,000 at a major tech firm is not going to jump at an offer of $80,000.”