Finding good data science talent starts with an effective job description. But to craft just the right posting, organizations must understand the talent market and role they are targeting.
In today's data science job market, demand far outstrips supply, says Chris Nicholson, co-founder and CEO of artificial intelligence and deep learning company Skymind, and co-creator of the open source framework Deeplearning4j. That means organizations must resist the temptation to seek candidates with every last required data science skill in favor of hiring for potential and then training on the job, he says.
"A lot of data science has to do with statistics, math and experimentation — so you're not necessarily looking for someone with a computer science or software engineering background, though they should have some programming experience," Nicholson says. "You want folks from physical science, math, physics, natural sciences backgrounds; people who are trained to think about statistical ideas and use computational tools. They need to have the ability to look at data and use tools to manipulate it, explore correlations and produce data models that make predictions."
Because a data scientist's job isn't to engineer entire systems, minimal programming experience is fine, Nicholson says. After all, most organizations can rely on software engineering, DevOps, or IT teams to build, manage and maintain infrastructure in support of data science efforts. Instead, strong data science candidates often have a background in science and should be proficient with data science tools in one or more different stacks.
Here's how to craft just the right data science job posting — and get talent in the door.
General best practices
When it comes to crafting any job description, you're primarily marketing your organization and the role, says Ammon Bartram, chief data officer and co-founder of Triplebyte. The goal is to communicate why that role is an exciting opportunity rather than focus only on the skills and responsibilities. A lot of organizations seeking data scientists make this mistake, Bartram says, putting them at a disadvantage right off the bat.
"Recruiters often write things like, 'Must have a technical degree, three years of experience, and deep knowledge of Apache Hadoop.' This is a mistake, even if you really want someone with these attributes," Bartram says. "For a high-skill role like data science, the goal is to convince applicants who might be on the fence that your company and your role are interesting and worth their time."
This is especially important not just because the market is so hot, but because, Nicholson says, "A lot of the necessary skills are industry- and company-specific. Organizations use different languages, prefer certain vendors' tech stacks and specific proprietary tools, so that is up to the hiring teams to know which ones."
Instead focus on the mission of your company, what the role will accomplish, and any technical details of the exciting problems candidates will get to solve, Bartram says. "For data science in particular, it can work great to write about the interesting data sets that the candidate will have access to — data science candidates love to geek out over cool data sets," he says.
By starting your job descriptions this way, you will more easily hook candidates, and then you can move on to the harder, technical skills necessary for the role.
Specific skills needed
Specific languages and tools will vary by company, but first and foremost is fluency in statistics, followed by some programming experience, and familiarity with data systems, Bartram says. Here, it's important to be specific where necessary.
"Beyond the basic skills, companies may want machine learning knowledge — understanding of specific ML models, and common tools to work with them. Tensorflow and scikit-learn are the two most common ML libraries in use," Bartram says. "Companies may want experience in whichever specific programming languages they use, like Python, Java, R, etc. And companies may want familiarity with the specific data systems they use, like PostgreSQL, MongoDB, Airflow, Hadoop, Redshift."
Nicholson adds Matlab, a numerical computing environment and a proprietary programming language from Mathworks, to that list of basics.
"These are tools that most natural sciences professionals and statisticians are familiar with, so that gives you a good feel for if candidates can think about these problems the 'right' way," Nicholson says. "And if they're working on machine learning — Python has dominated as the language of choice in this field, as well as Python tools like Canvas, Keras and Tensorflow. If they have one of those but not the other, that's not a deal-breaker, though. It's more important that they're adept at one, because that shows they can pick up new tools quickly."
Pavel Dmitriev, vice president of data science at sales engagement platform company Outreach, uses a more extensive list when hiring, adding that, to be successful at Outreach, candidates should have working knowledge of all these but be extremely proficient in at least a couple.
"Some of this is domain-specific; for us, natural language is very critical because of our business, but in other companies it might be something different," Dimitriev says. "But, in general, we look for skill with coding and algorithms; data manipulation; big data management; machine learning; natural language processing; business understanding; how to formalize problems and transform them into mathematical problems; and communication skills."
Because data scientists must collaborate with a wide array of colleagues, soft skills are essential, and should not be overlooked in any job description, says Bartram. Communication, teamwork, collaboration as well as passion and mission are critical for any data science candidate.
"Soft skills matter," Bartram says. "Being a data scientist involves communicating with co-workers and working on a team. The main two skills here that impact whether a candidate receives offers are their communication ability, and their enthusiasm for what the company is doing."
Be sure to include not only your preferred soft skills for the role but also a brief explanation of how data science fits into your organization — that way candidates will have a better idea of how those soft skills will be put to work in the role.
Now that you have all the pieces, you can put together a stellar data science job description that's sure to land great candidates.