How companies are creating data scientists in-house

Ten years ago, companies were buzzing about big data, but few organizations beyond giants such as Amazon, Netflix and Google actively recruited IT pros with deep data chops. That has changed in the past few years thanks to improved tools and training, as well as evidence that data can drive business value. Now data science skills are in high demand in nearly every industry.

A recent query of job search site Monster.com showed over a thousand full-time data science postings from employers spanning the gamut from government and healthcare, to consulting firms and software companies. Demand is so great that it is unlikely there are enough long-tenured data scientists or math Ph.D.s to go around.

Plus, identifying — and validating — data science talent is complicated by the emerging nature of the discipline, says Mark Jacobsohn, senior vice president and analytics lead at Booz Allen Hamilton.

“Unlike other professions, we have not seen a single widely recognized certification for data science. There is disagreement on what the term means. That makes recruiting and training more challenging,” Jacobsohn says.

[ Keep up to date with the 10 hottest data analytics trends today (and 5 going cold). | Bolster your career with our guide to the big data certifications that will pay off. | Get the latest on data analytics by signing up for our CIO newsletter. ]

Because of this, many organizations are turning inward to fill their data needs, establishing training programs to gear up current employees with the latest tools and techniques of data science.

Here is a look at how several companies are developing data science talent.

Identify where data science will add the most value to your organization

Before establishing a data science talent and development program, it is worth ascertaining where such an effort will pay the most dividends. Because data science blends analytics with business decisions, much can be gained by targeting employees with domain expertise, in addition to technical promise. For many organizations, the best use case for data science to add business value remains marketing and technology platforms with high activity levels.

“Our latest recommender algorithm drove a statistically significant lift in the number of engagements within our application. When expanded to our entire user base, this will translate into millions of dollars of incremental bottom line revenue at Ibotta,” says Bijal Shah, vice president of data products and analytics at Ibotta, a software company that allows consumers to earn cash back on purchases made through an app.

Outside of sales and marketing, data science and analytics add value by improving productivity. For example, GE has worked with steel companies to improve the efficiency of high-value production equipment. In the steel production case, data is obtained from sensors attached to high-value production equipment. Data science analysis can then make predictions on the best time to apply preventive maintenance to help avoid unplanned downtime.

For Booz Allen Hamilton, which is working to equip hundreds of its staff with data science skills over the next few years, special industry considerations have been a driver for internal development of data science skills at the firm.

“Locating professionals with the appropriate security clearance is a challenge in our business because we work with the government. That is one of the reasons why we are developing our existing staff,” Jacobsohn explains.

Equip everyone with data literacy

For some organizations, data science development is an across-the-organization, cultural affair.

“I am working on developing a culture of data literacy at Qlik,” says Jordan Morrow, data literacy program manager at Qlik, a data analytics platform maker. “Our training and development approach offers beginner, intermediate and advanced training for data science. In our training, the critical turning point is changing the perspective on data. Many people view data as reporting or simple summary statistics. We want to equip our staff to ask deeper questions with data.”

This data literacy approach may represent the long-term future for data science. As AI and other tools continue to improve, more people will have the ability to ask data-related questions. When spreadsheet software was invented, did anyone imagine that non-accountants would use it? For data science, the day may come when improved tools and widespread education give everyone data scientist capabilities.

Equipping staff to deliver data science results

The most popular data science tools remain Python and R for many organizations. For those tools to be useful, staff need a foundation first. “I went through the firm’s two-week data science training program which equipped me with a variety of skills including using Python and machine learning,” says Brad Morgart, associate at Booz Allen Hamilton’s analytics group. “In the past, I worked extensively with data using Excel and Access. The fundamentals of statistics and analysis were not new to me from my studies in economics and business. However, I am now able to analyze much larger amounts of data more quickly in my work.”

Morgart’s success shows that the data science experts of tomorrow will not come from computer science and math programs alone. Booz Allen’s type of foundation training is important because providing data software tools alone cannot develop good data questions or connect data science back to business needs.

Developing a two-week training program is just one part of Booz Allen Hamilton’s commitment to data science. “We are investing heavily in data science including training and recruiting. There is significant demand for these skills from our clients. We also provide additional online courses and support mentorships to help our staff grow,” Jacobsohn adds.

No capacity for an internal program? Partner up

Developing an internal data science education program may not be right for every organization. Here, it may make sense to partner with an external organization to help develop in-house talent.

“Over 400 AT&T employees have enrolled in the Georgia Tech Online Master of Science in Computer Science program. The program has produced over 500 graduates, including nearly 50 AT&T employees. In addition, we worked with Udacity on Nanodegrees — self-paced, fast-track technical credentials in areas like machine learning, artificial intelligence, and data analytics. Over 2,000 AT&T employees are among the over 30,000 learners worldwide who have enrolled in Nanodegree programs,” explains John Palmer, senior vice president of human resources and chief learning officer at AT&T.

In addition to developing current employees, AT&T has looked at these programs as a source for new talent. “We committed to hiring 100 Udacity Nanodegree graduates as interns, and have hired three classes of summer interns, 2015-2017. We’ve offered many of them full-time positions,” Palmer says.

Online education programs for data science and related programs are maturing. They are a viable way for individuals and companies to acquire new capabilities.

The role of manager support in growing data science talent

Once your staff have acquired skills, the next step is providing an opportunity for them to practice. “I asked the head of analytics, Bijal Shah, to [allow me to] work on ‘Project Churn,’ a project dedicated to identifying key indicators of user retention via predictive modeling. Despite being the most junior member on the team, Bijal took a risk and allowed me to play a significant role in the project. This not only inspired confidence in myself, but challenged me to build new skills, both technically and strategically,” says Charley Frazier, senior analyst of data products at Ibotta.

“One of the most critical components of breaking into data science was having managers and mentors who empowered me to learn new skills, take on new challenges, and not be afraid to fail and learn from my mistakes,” Frazier adds. Setting some time aside for learning and experimentation is a critical way to support the growth of data science.

Pursuing development opportunities outside the office

Learning on the job may not be enough for professionals who want to advance. These employees should be encouraged to grow their skills outside the office as well — especially through online courses and in-person conferences.

“I spent considerable time outside of work learning Python programming and machine learning foundations,” Frazier says.

For Frazier, online courses stood out as particularly valuable ways to acquire skills.

“Colleagues across different teams also pair up and complete MOOC [massive open online courses] courses like Andrew Ng’s Machine Learning Coursera class,” says Frazier. “Ibotta encourages and funds members of the technology teams to attend a conference of their choosing each year,” Frazier adds.

Data science events and conferences vary widely in their style and emphasis. The Strata Data Conference, presented by Cloudera and O’Reilly Media, features presentations from Microsoft and GE Digital as well as technical presentations on various software applications. Other data science events deliberately emphasize a practitioner focus and exclude sponsors. The Data Science Conference, which recently included speakers from Amazon, Allstate, Deloitte, eBay, Facebook and Microsoft, is an example of that approach. As data science is still an emerging discipline, enabling your employees to participate in conferences remains an important way to help them develop new skills — and recruit new employees.

The forecast is improving for data science recruitment

In recent years, recruiting data science professionals was difficult. For example, Microsoft recruiters looked at conferences attended by specialized researchers to find talent. This approach succeeds if your organization can offer compelling opportunities suitable to attract PhDs. What if you do not have that type of research-oriented environment?

“Over the past few years, we are seeing an increasing number of undergraduate and graduate programs in data science. In fact, some of our interns in 2017 are bringing great data science skills to the table. It is possible that we may be able to recruit from such programs for our needs in a few years. In the meantime, our approach balances external recruiting with developing internal staff,” Jacobsohn says.

Related data analytics articles: