Data scientists find lower salaries, greater fulfillment in public sector


John Farmer, the director of technology and civic innovation at Microsoft New York, co-leads a panel on how to use data to make communities more resilient during disasters at Bloomberg's 2015 Data for Good Exchange.

John Farmer, the director of technology and civic innovation at Microsoft New York, co-leads a panel on how to use data to make communities more resilient during disasters at Bloomberg’s 2015 Data for Good Exchange. Photo: Wendy Lu.

More than 400 data scientists, nonprofit workers, academic scholars and industry experts strode through the glass doors of Bloomberg Tower recently to attend Bloomberg’s second annual Data for Good Exchange, a day-long conference that aims to connect data-driven professionals across disciplines.

The conference explored a new trend of technologists who are increasingly interested in using their data skills to tackle social issues such as poverty and sex trafficking instead of – or in addition to – working for Fortune 500 companies.

“Coming out of graduate school, I realized the data revolution was changing the way humans use information. That felt so powerful, and yet all of the places we’re getting really good at using this data were in Silicon Valley and Wall Street,” said Jake Porway, executive director and founder of DataKind, a 3-year-old nonprofit organization that connects governments, nonprofit groups and for-profit organizations with data scientists who work pro bono to create digital solutions. “I recognized there are all these opportunities to use the same algorithms that help companies boost profit to instead help social organizations boost their impact.”

The idea for DataKind started out as a simple post on Porway’s blog in the fall of 2011, asking friends whether they’d be interested in partaking in a “data without borders” initiative to match data experts with socially conscious companies. The blog post went viral: Suddenly, Porway was receiving tweets from Tim O’Reilly, and even the White House featured him at their events.

Headquartered on 156 5th Avenue in New York, DataKind is now comprised of a full-time staff of 11 and nearly 7,000 volunteers working in six chapters around the world, including Singapore, Bangalore and the United Kingdom.

Porway, 32, studied computer science as an undergraduate and received a postdoctoral degree in statistics from the University of California in Los Angeles in 2010.

“Especially in the data science profession, it can be very easy to make a lot of money doing this work, but not feel like you’re giving much back to the world,” said Porway, who co-led a panel at Data for Good Exchange on the use of data for nonprofits. “People want to give back, but it’s hard to just call up a nonprofit and say, ‘Hey, I can build you statistical models — want it?’ Most people wouldn’t even know what to do with that.”

He added that nonprofits have trouble raising funds to hire data scientists because most organizations either don’t understand the value of data science or, even if they do, are unsure of how to hire or look for the right people. At DataKind, many of their nonprofit partners end up hiring from the pool of volunteers because of the projects they worked on for them.

In the winter of 2014, DataKind matched a principal data scientist from Pivotal Software, a for-profit company in Palo Alto, Calif., with Crisis Text Line, a Manhattan-based nonprofit that provides free 24/7 crisis counseling via text message.

At the time, only 3 percent of the nonprofit’s texting clients were taking up 34 percent of the crisis counselors’ time, said Bob Filbin, chief data scientist at Crisis Text Line. The problem was that many texting clients returned to Crisis Text Line for additional help, misusing the counselors as a replacement for long-term therapy.

For two months, Filbin worked closely with Noelle Sio, principal data scientist from Pivotal, to identify at what point Crisis Text Line should provide returning clients with appropriate long-term care resources.

“We’re talking about shaving off 15 conversations per texter. Those 15 conversations per [returning] texter can be reallocated to new texters in crisis,” said Filbin.

Consequently, Crisis Text Line will be able to help 24,000 more texters in 2015 and an additional 100,000 texters in 2016.

For a partnership to be successful, Filbin added that it’s crucial for the data scientists to understand the nonprofit’s mission and the context for the data they’re analyzing.

Nonprofits like DataKind and Crisis Text Line aren’t the only organizations interested in leveraging technology for society’s benefit. For the past two months, Microsoft New York has been working with Ushahidi, a global technology nonprofit based in Nairobi, Kenya, to provide internet for communities that experience low connectivity during disasters.

For instance, when Hurricane Sandy led to $19 billion in damages across New York in 2012 and people were out of power for days, many of the challenges brought on by loss of electricity required data, according to John Farmer, the director of technology and civic innovation at Microsoft New York, located in Times Square. Providing go-to technology tools as simple as Microsoft Excel, he said, could reduce lag time for first responders trying to rescue victims.

“We want to speed up the pace, assess damages on the block and get information in by 10 a.m. instead of 8 p.m.,” he said, during a Data for Good Exchange panel on using data to make communities more resilient to disasters.

The project will begin testing its product in communities in New York and Nairobi by simulating disaster-like environments without connectivity. They are still working on finding partners, and would ideally like to work with first responders such as the New York Police Department or city executives, as well as humanitarian organizations like Red Cross.

Nathaniel Manning, the director of business operations at Ushahidi, said the nonprofit has hundreds of volunteers and only a few full-time paid staff members. Depending on whether Ushahidi staff members are based in Kenya or the United States, their salaries range from $40,000 to $80,000.

“We work with what we have, and we’re able to write our own hours. It’s all driven by metrics and not by the amount of time you sit in the office. We’re pretty competitive and we’re able to pay people,” Manning said.

Hannah Fresques, 28, said if she had to choose between a high-paying job and working at a nonprofit with a mission statement that aligns with her values, she would choose the latter.

“Social good is important to me in terms of basic values of fairness and trust,” said Fresques, who recently graduated from Columbia University with a degree in quantitative methods for social science. She hopes to find a job as a data analyst at a nonprofit, preferably in New York.

Experienced data scientists who are looking to leave the private sector, where they are used to six-figure salaries, to work in the public sector are typically offered a salary of around $50,000, according to Magdalena Gyerko, managing partner of information technology recruiting at Lucas Group, an executive search firm on 500 5th Avenue.

As a result, companies in the public sector tend to be more flexible in their hiring policies because their compensation is lower. Typically, data scientists looking to move to the public sector may do so for the sake of work-life balance.

“The trade-off might be the change in lifestyle. They’re giving up financially on a few thousands of dollars, but in return, they’re not working 80 hours a week,” Gyerko said.

Porway agreed that data scientists don’t get the usual monetary benefits at a nonprofit that they would at a for-profit company.

“That’s the problem with the nonprofit model — it’s hard to get funding for anything, and when you do there’s this stigma that you shouldn’t be paying yourself as much as the market because that’s dollars that could be going to the mission,” Porway said.

Matt Ernest, a senior data analyst at Mediacom, said he hopes to move from working for a corporate client to using his data skill set full-time for a cause he cares about.

“A lot of research takes five to 10 years to do, and policy changes take another five to 10 years,” he said. “That’s helpful, but it doesn’t help people immediately. Digital data is a way to do that.”

“You can’t do everything in life expecting to get paid for it. In the business world, the more you want to do something that’s not beneficial to corporations, the more difficult it’s going to be,” he said.