In the early days of computing, developers were often jacks of all trades, handling virtually any task needed for software to get made. As the field matured, jobs grew more specialized. Now we're seeing a similar pattern in a brand-new domain: big data.
That's according to P.K. Agarwal, regional dean and CEO of Northeastern University's recently formed Silicon Valley campus, who says big-data professionals so far have commonly handled everything from data cleaning to analytics, and from Hadoop to Apache Spark.
"It's like medicine," said Agarwal, who at one time was California's CTO under former Governor Arnold Schwarzenegger. "You start to get specialties."
That brings us to today's data-scientist shortage. Highly trained data scientists are now in acute demand as organizations awash in data look for meaning in all those petabytes. In part as a response to this, other professionals are learning the skills to answer at least some of those questions for themselves, earning the informal title of citizen data scientist.
"The profession is subdividing," Agarwal said.
The result is that there's now a multipronged approach to both tools and education.
"It used to be that anybody who was in the business world needed to learn PowerPoint and Excel," Agarwal said. "Probably five years from now, Microsoft Office will have something combining Excel, R and Tableau. It's just natural. If there's this new class of citizen data analysts, they're going to need new tools."
They can learn these skills in a variety of formats and places, and with varying areas of focus. Even as hard-core "quants" are acquiring ever more sophisticated skills to help meet the demand at one end of the spectrum, traditional business people are getting more savvy about data analysis and presentation.
Northeastern offers eight-week "bootcamps" in data analytics aimed at a broad spectrum of business people, as well as more intensive certificate programs for professional data scientists. Specialized master's degree programs are also in the works.
"My sense is that it will take two to three years to even out the shortage, and tools will become much more sophisticated," Agarwal said. "We're still not automating things as much as we could."