Many missions require disconnected and portable compute capabilities that enable teams to execute time-constrained operations that process high volumes of enterprise data. However, the time-intensive, computationally expensive nature of these data ingest, data processing, and AI/machine learning (ML) activities routinely hinder organizations from achieving desired mission effects. While technological advancements in small form-factor devices can provide missions with the necessary compute resources, challenges remain.
For example, operators in the field often lack engineering proficiency to build high-performance computing (HPC) software. In addition, integrating and configuring these HPC systems with advance compute necessitates specific expertise and often requires custom or bespoke implementation that limits modularity, portability, and scalability.
As a result, mission applications fail to scale, execute too slowly, or cannot be executed. A lack of simple workflows prevents field operators from integrating HPC software and achieving a desired mission effect with portable, advance compute. Expressly to address these challenges, we built Booz Allen Precog™, a general-purpose software application with a low-code interface that compiles reusable data pipelines into graphics processing unit (GPU)-enabled software.
