GTC-DC 2019: Big Data Platform: The Killer DevSecOps Platform for Federal AI Projects (Presented by Alion Science)
John Eubank, Alion
We’ll focus on the Department of Defense’s Big Data Platform (BDP), which is being used for multi-domain operations. Alion’s approach to DevSecOps aligns with the DoD’s DevSecOps Reference Design to deliver capabilities rapidly by working across infrastructures to obtain resource efficiencies between the cloud and on-premise infrastructures. In Alion’s implementation of DevSecOps, the BDP is the cohesive intermediary where data is securely stored, normalized, labeled, and made available to internal and external services. It’s now the go-to architecture for massive scale data processing in which AI applications can be rapidly developed, trained, and deployed. The BDP ecosystem allocates between GPU and CPU nodes to manage the compute pipeline to optimize the workload’s infrastructure resource usage and processing performance. A government open source architecture, the BDP is available to the U.S. federal government.