AI Factory Accelerating Solutions with Ray

The Lockheed Martin AI Factory is a secure end-to-end modular ecosystem to train, deploy, and sustain Artificial Intelligence solutions. The AI Factory is founded on a set of centrally hosted cloud and bare metal environments including edge. This enables users to scale, train, monitor, and serve their intelligent systems.

Machine Learning (ML) use cases benefit from Ray's ability to distribute workloads across AI Factory's platform. Using Ray Clusters, large object detection models, such as YoloV5, have been trained, tuned, and are ready to serve. Additionally, we are working to integrate the full Ray ecosystem; including Ray Spark, KServe, and Ray AIR. Through AI Factory's Open-Source contributions, we ensure network traffic is authorized and authenticated per request. Prior to these enhancements, a set of Software Patches were maintained internally and installed alongside Ray.

A key feature of AI Factory is the ability for programs to independently manage their Ray Clusters, minimizing barriers to entry. This system incorporates Cloud Native Computing Foundation concepts, enabling the highest security standards and reliance. A focus on API First design, Reusable Architectures, and User Focus are core principles during tool assessments. During last year's Ray Summit 2022, feedback was provided to the Ray community improving alignment with our principles. These topics will provide value to the Ray community through the lens of a large organization.