Data Science Talent Shifts
The Talent, the New Recruiting Rules, and the Long-game
I’ve worked in the data science and analytics industry for ten years and have spent a lot of that time overseeing talent acquisition and development. But over the last five years or so, I see our profession shifting. No longer is business intelligence reporting enough. As more standard reporting and dashboarding can be automated, the next frontier is personalizing the customer experience end-to-end. And that means a change in focus, from descriptive analytics to predictive science.
To advance their analytics, companies like us need to heavily invest in data science tools and talent. But even as data science becomes more important in the economy, many things are changing: the technology, the skillsets required for the job, and the opportunities that data scientists have available. Adjusting to this new world can be a challenge for established analytics firms.
Talent Is Changing
Over the last decade, I’ve seen the archetypical data science profile shift from jack-of-all-trades toward specialist. The range of programming languages, technologies, methodologies, and domain knowledge is too broad. We can no longer expect our talent to simply “wear many hats”, being skilled in every piece of our business; we need people who can go deep on certain subjects, and rely on their peers to be experts in other domains.
So how has 84.51° grappled with this? We’ve begun to think of our science talent in three profiles, or specialties:
Insights specialists are masters of getting information out of data. They often bring strong business acumen to the table, along with great communication skills and a gift for visualizing their findings for a non-technical audience.
Statistical learning specialists are uniquely knowledgeable about cutting-edge techniques and applications. Scientists with this specialty are our go-to experts for statistics, model development, and machine learning.
Technologists tend to come from a background closer to computer science. They are experts in building tools (like web apps, containerized services) and automating tasks. They are also likely to be the early adopters of new technologies that we bring in-house.
Articulating these three types of data scientist has helped us think more clearly through our hiring process. It has also helped us better classify the needs we have within our business. For example, if one of our teams is trying to fill a role, it’s helpful to know they’re looking for a technologist specifically (versus a generic data scientist).
Talent Is Hard to Find
Not only is talent becoming differentiated, it’s also becoming more difficult to find. It’s no secret that companies are investing heavily in data science, and demand for data scientists has skyrocketed in recent years. The supply just hasn’t kept up, and potential candidates are highly sought-after, providing them many available opportunities. But those candidates are looking for the right work – work that is challenging and intriguing to them. It’s become more important that we advertise the engaging, meaningful work we do: building new sciences, doing deep dives on what draws people to Kroger. Additionally, location matters, and we’ve invested in building out a “Tech Hub” in Chicago. Adding geographic flexibility, allowing data scientists to live in a major metropolitan area, has widened our recruiting pool.
This scarcity of talent has also made it expensive to recruit experienced data scientists. At 84.51˚, we’re very confident in our ability to develop talent in-house, so we rely heavily on our Data Science Development Program. We seek out young data scientists with a strong foundation and invest in them, building their skills over an 8-week onboarding program and then with real project work that they do with guidance from technical mentors.
New Recruiting Rules
In the past, we could let candidates come to us. Our talent pool – particularly those coming out of a university program – had fewer choices of employer. Now, however, we need to actively seek out the right data scientists.
So just as applicants submit resumes to us, we offer our own resume to candidates: this blog, our scientists’ appearances at conferences, and even our Area 51 data sets. We want to show that our work is interesting and our company is a great place to work. Publicizing ourselves is particularly important because we’re expanding our reach; in the past we were able to fill our roles largely with local talent, but that’s no longer the case.
These new challenges aren’t just limited to finding talent, and the industry hasn’t stopped changing. We know that the ground is constantly shifting under our feet. Keeping up is hard, but we can lean on our day-to-day practioners. Our data scientists already need to be in touch with what’s happening in the field, so their opinions inform our vision of what’s most important in recruiting.
The Long Game
We recruit learners – people who are excited about the work and want to be part of a community. But exit opportunities are plentiful for data scientists, and if we don’t provide the environment we promised, all of our hard work to get them in the door is wasted.
What exactly does this look like? The particulars are always evolving, but we focus on a few things. To build community, we regularly host social events and continue to create an open, friendly culture. To help our people keep learning, we invest in online resources (like Safari Books Online) and sponsor hack days, both internal and external. We also bring in third-party trainers (like Anaconda) to give our data scientists an opportunity for in-person education, and we encourage our internal experts to deliver trainings themselves. To keep the work interesting and fast-paced, we’re careful to place our specialists in roles that fit their skills but also provide them opportunities to branch out.
Approaching recruiting thoughtfully, taking all these things into account, has really changed 84.51˚ for the better. The environment will continue to shift, and we’ll remain attentive. But we also know that nothing else matters if we forget the important things: maintaining a fun work culture, doing interesting work, and treating our people well.