As more organizations become aware of the central role data plays in their business processes, there's more demand for skilled workers to handle various data management tasks. But there's also more confusion around the differences between positions like data architect, data modeler and data engineer, and which data management roles are most valuable to an organization.
Michael Bowers, chief architect at NoSQL database vendor FairCom Corp., sought to cut through some of that confusion during a session at Dataversity's inaugural Data Architecture Online event. The online conference focused on key strategies and technologies that are needed in order to build and manage a modern data architecture.
In his session, Bowers, who has more than 30 years of experience as a data professional, compared eight data management and analytics jobs in particular, including their key functions, their salaries and the technical skills they require. He also offered advice on how to increase one's salary in the data management field and how organizations should go about building a data management team.
The job descriptions and salaries, which are averages from across the U.S., come from job sites glassdoor.com, indeed.com and payscale.com, as well as Bowers' own work experience. He noted that he has held, managed or led projects involving all of the data management roles he discussed except for data scientist.
This article is part of
The differences between data management jobs
The positions bringing more value to the business -- like data architects, data modelers and data scientists -- are harder to fill, Bowers said. That's because people in those positions generally need to be more familiar with cutting-edge technology than other data management workers are.
How do these data management roles compare? Data architects design and help implement database systems and other repositories for corporate data, Bowers said. They're also responsible for ensuring that organizations comply with internal and external regulations on data, and for evaluating and recommending new technologies, he added.
Bowers described a data architect as a "know-it-all" who has to be familiar with different databases and other data management tools, as well as use cases, technology costs and limitations, and industry trends. "I had to master a ton of technologies to become a data architect," he said.
A data modeler identifies business rules and entities in data sets and designs data models for databases and other systems to help reduce data redundancy and improve data integration, according to Bowers. Data modelers make less money on average than many other IT workers, but you get what you pay for, he cautioned.
"It's hard to find a good data modeler," Bowers said. "It's an art. Anyone who says you can just throw data into a computer and get a good model out automatically is wrong."
Software engineers who are also good at data modeling can deliver the best results, he added, though it typically is more expensive to hire them.
Big data focus for data engineers, data scientists
A data engineer is essentially a BI engineer for big data, Bowers said. Data engineers build data pipelines connecting databases and big data systems, and like data architects, they must understand the intricacies of various cloud and on-premises technologies.
But in dealing with big data, data engineers can often be distracted by those technologies rather than focus on delivering business value, Bowers said. "They just spend all their time playing and not getting [things] done."
Meanwhile, data scientists -- who find, prepare and analyze data using machine learning algorithms and other advanced analytics applications -- can deliver problematic results without proper knowledge and application of statistical principles, which Bowers described as the foundation of data science.
Developing, testing and validating predictive models "really does take a Ph.D.-level statistician," he said. "It's really easy for people to claim they're data scientists and not know anything about statistics."
Overlapping skills in data management positions
Data management roles can also easily overlap on skills, Bowers said. For example, in larger enterprises, database administrators (DBAs) often are specialized to the point where they just develop, operate and administer databases. But in many small companies, a DBA is more of a "catch-all" employee who also does data quality, modeling, analysis and reporting.
Nonetheless, Bowers said DBAs are often undervalued for the work they do. At an average base salary of about $81,000, a typical DBA makes less than all the other data management positions he discussed except for a data quality engineer ($74,000) and a data analyst ($67,000).
"The skill level required to be a DBA, especially an Oracle DBA, is so high that [their salary level] is just insulting to me," Bowers said. "It really bothers me they are making so little compared to what it takes to become a DBA."
However, DBAs who focus only on operational tasks are putting themselves in a tough position, Bowers warned. Such DBAs have been around a long time, so they are easier to find, but he said their business value is lessening as automated database administration, driven by machine learning and advanced analytics, becomes more prevalent.
"They do things by hand, and they aren't pushing themselves to do new things," Bowers said. "If you're an operational DBA, move away from that as quickly as possible and toward modeling or development or BI engineering."
Top salaries and how to boost them
The highest average base salary Bowers cited is $124,000 for a data architect. That's followed by a data scientist and a data engineer at $117,000, a BI engineer at $106,000 and a data modeler at $91,000.
These salaries differ based partly on a position's value to the company. For example, Bowers said data engineers and BI engineers have similar functions, but data engineers will make around $10,000 more because of their greater familiarity with new technologies and processes.
Bowers concluded his pay analysis by explaining how data professionals can increase their salaries. His first piece of advice: have a deep knowledge of a wide variety of technologies. Statistics, machine learning and new technical skills are particularly valuable, he said. Working more closely with IT managers is also beneficial, Bowers added. He said data architects make more money than the other positions partly because they've gained the trust of IT management through direct interactions.
The key, Bowers said, is to push yourself to master whatever technology comes your way.
"Know it inside and out, apply it and then move on to something else," he said. "You'll be amazed over the decades what you can master and accomplish. But you can't do it working 40 hours a week. You can work for your company 40 hours a week, but you should spend 10 to 20 hours a week mastering things beyond your current job."
Which positions to fill first
In building a data management team, some positions should take priority over others, Bowers said when wrapping up his conference session. The "go-to positions" he thinks organizations should fill -- with workers straight out of college, if possible -- are data analysts, BI engineers and data quality engineers. DBAs are also a must, he said.
By relying on them to do much of the hands-on work, you can minimize the number of data engineers, data scientists and data architects you have to hire, Bowers said, describing that as the most cost-effective way to staff up on the various data management roles.
But you need to find "really good people" for the latter positions, he cautioned. Otherwise, you could end up filling those key jobs with employees who "can talk circles around you and you don't even know that they're just blowing smoke at you," Bowers said. "You've got to hire people that you trust and who know what they're talking about."