LIFT AND SHIFT OF MODEL CODE USING MACHINE LEARNING MICROSERVICES WITH GENERATIVE AI MAPPING LAYER IN ENTERPRISE SAAS APPLICATIONS

In traditional Software as a Service (SaaS) enterprise applications, there is a need for easy-to-do machine learning (ML) frameworks. Additionally, SaaS applications are closely related when they form an application suite, which brings forth the need for an ML framework that can facilitate the “lift and shift” of ML model code in similar needs in multiple enterprise applications in a suite. To add to this, some SaaS applications are still using legacy infrastructure (on-premise) mandating the need for an ML framework that is backward compatible with coexisting platforms, both cloud and legacy on-premise infrastructure. This study first demonstrated that in SaaS applications, microservices were important ingredients to deploying machine learning (ML) models successfully. In general, microservices can result in efficiencies in software service design, development, and delivery. As they become ubiquitous in the redesign of monolithic software, with the addition of machine learning, the traditional SaaS applications are also becoming increasingly intelligent. Next, the dissertation recommends a portable ML microservice framework Minerva (also known as contAIn—second generation), a Micro-services-based container framework for Applied Machine learning as an efficient way to modularize and deploy intelligent microservices in both traditional “legacy” SaaS application suite and cloud, especially in the enterprise domain. The study also identified and discussed the needs, challenges, and architecture to incorporate ML microservices in such applications. Secondly, the study further identifies that there is an impetus to innovate quickly for machine learning features in enterprise SaaS applications. Minerva’s design for optimal integration with legacy and cloud applications using microservices architecture leveraging lightweight infrastructure accelerates deploying ML models in such applications. The study highlights the real-world implementation of Minerva, doubling innovation speed with the human resources. It evaluates ML model code reusability across applications, resulting in 1.15 to 2X faster adoption compared to previous methods in a marketing application suite. Minerva’s top-tier security encompasses several advanced features designed to protect sensitive data in SaaS marketing applications. It includes end-to-end data encryption, ensuring all data remains secure both in transit and at rest using robust cryptographic algorithms. While a layered design accelerated innovation through porting existing models to related business suites, generative AI methods, while promising, hadn't yielded significant gains with smaller models yet porting over already no code optimized model code.