DISA looks to pilot AI platform, DevSecOps model by summer

The Defense Information Systems Agency is building an artificial intelligence platform as a service to facilitate clearance applications being checked and verified against multiple data sources. DISA plans a pilot by June.

Terry Carpenter, DISA's program executive officer for the National Background Investigation Service, told FCW Jan. 24 that the agency is actively looking for algorithms and data sources to improve the NBIS and plans to use it to help improve DISA's own product, which currently can handle only simple algorithms.

"If the system sees an anomaly, it flags it and sends it up for a human review," he said. "We can add artificial intelligence and machine learning to say not only is this an anomaly but here's a risk assessment of the anomaly, the data used to come up with that and the mitigation recommended."

DISA also has been working to aggregate and funnel the data sources to be used for background investigations. "We're bringing them into one security enterprise architecture data broker function that is going to supply data to the whole system," Carpenter said.

Additionally, the agency is working on a DevSecOps model for NBIS that is expected to launch this summer.

The main challenge in getting the DevSecOps model to work, Carpenter said, has been the way the NBIS program was structured with different pieces built by different partners, spread out across multiple clouds with their own tools and processes and the rest in the data center.

"We're still releasing in a rapid fashion, I would say we're close to a devops model but we're not fully there because we have so many clouds now and different tool sets," Carpenter said.