How federal agencies approach data has a lot to do with mission success, according to veteran federal IT experts who took part in a recent Association for Federal Information Resources Management (AFFIRM) panel discussion hosted at the CGI Innovation Center in Arlington, Va.
Data can enable powerful government services, but an array of challenges often inhibit agencies’ ability to leverage data as a true strategic asset. Cultural, regulatory and technical issues all create hurdles, the panelists agreed.
Gundeep Ahluwalia (CIO at the Department of Labor) and Melvin Brown II (Deputy CIO at the Office of Personnel Management), joined by CGI Director Rajesh Vasisht, were the panelists.
Data sharing—an imperative with inherent challenges
Data modernization and sharing is the foundation of mission fulfillment, Vasisht said. Agencies have mountains of data that are traditionally isolated, but that they are moving towards a mission focus that is enabled by sharing data across the federal government, he added.
At DOL, Ahluwalia said missions that will be impacted include facilitating unemployment claims and transitioning military retirees into the commercial workforce. Brown noted that at OPM, personnel information for 8 million federal employees and retirees resides in their various data stores. “That’s a massive amount of data,” he said. OPM built a web-based dashboard that allows agencies to see their employees’ data, thereby enabling their recruiting missions.
Ahluwalia reinforced the criticality of cross-organizational data sharing to achieving mission outcomes that positively impact Americans. He noted that his agency gets the DD214 form—discharge papers for active-duty military personnel leaving the service. This form includes the person’s discharge location, date of discharge and skills. “I should be able to serve them 50 (job) opportunities in their zip code that are relevant to them” he said, which requires combining data on the military member with employment opportunity information and other data from various organizations.
But sharing data across agency boundaries is not always easy. Ahluwalia said regulations often pose his highest hurdle. Regulations that prevent data sharing in the interest of protecting privacy, for example, serve an important purpose—however, sharing data between agencies is often necessary to achieve the agency’s mission. Looking forward, Ahluwalia cited the need for regulations to better balance security and privacy with function.
Cultural challenges also persist, as some organizations that own data resist co-mingling it with other organizations’ data sets, he added. The proverbial silos arise in many cases because data owners are reluctant to open their information to other entities.
Shaping tomorrow’s data workforce
Reskilling and upskilling existing employees are essential for agencies to adapt to changing data structures, uses and regulations, the panelists agreed. Data scientists are in high demand, and the government competes with private industry to hire them. Crowdsourcing can provide help in some instances, when all the data involved is suitable for public release, Ahluwalia said. “If you throw your data out there, the data scientists that are out there will give you insights.”
Internship programs are another rich vein for mining talented data scientists, Brown said. OPM has taken a new approach to their training: Rather than pairing the new interns with old hands for mentoring, interns get problems to solve early on in their tenure. “Because they don’t know what they ‘can’t’ do, they’re solving the problems” in innovative ways, he said.
Another key point: Hire new employees to move the agency forward. “We’re not just backfilling positions,” Brown said. “We’re hiring for the future.”
Embracing the cloud to evolve data as a strategic asset
The logistics of storing and retrieving data has evolved significantly—from mainframes to the cloud—over the decades, Vasisht said.
“Before the cloud, there was mainframe. Everything was in a big box. What if the big box breaks? You have to buy another big box,” he said. The obvious cost and architectural constraints of this arrangement were always a concern, but for many years, technology offered no better options.
When cloud computing finally became powerful and secure enough for the government to embrace it, agencies began moving data from on-premises storage into the cloud. Initially, it was simply a lift-and-shift, moving data from the agency’s data center to the cloud provider’s platform, but eventually agencies began to reimagine the future state of the data.
Today, tools provided by cloud providers enable agencies to migrate tremendous amounts of data securely into cloud-hosted “data lakes” and make it available as needed, Vasisht explained. Making the data broadly accessible with minimal constraints and enabling secure retrieval create the potential for powerful new insights.
The promise of emerging technologies
Artificial intelligence and machine learning are already part of the IT picture, and rapidly becoming more mature. These emerging technologies have tremendous potential to make data sharing both easier and more impactful, Ahluwalia said.
At Labor, he said, the department’s Occupational Safety and Health Administration (OSHA) takes information from employers on workplace injuries and must code the reported information correctly to enable workers’ compensation claims. Through machine learning, OSHA can now code about 96% of the reports automatically, he said.
AI and ML are increasingly becoming commonplace elements of solutions developed by CGI, Vasisht said, noting that the firm has made strategic investments in emerging technologies including AI, graph databases and the metaverse.
Data is the key enabler for all phases of the machine learning pipeline, from data collection, curation, and exploration to model development, training, deployment and monitoring. All of this suggests that data modernization advances happening now are set to pay off soon, Brown said. “Five years from now, the focus needs to be on how we get value out of the things we’ve done,” he added.