A Comparative Study on AI/ML Optimization Strategies within DevOps Pipelines Deployed on Serverless Architectures in AWS Cloud Platforms

Abstract

The application of Artificial Intelligence (AI) and Machine Learning (ML) in modern DevOps pipelines is a rapidly growing trend, with organizations seeking efficient, scalable, and cost-effective solutions to integrate AI/ML models into production environments. AWS's serverless architecture, with its powerful cloud-native services such as AWS Lambda, Step Functions, and SageMaker, provides a flexible platform for deploying AI/ML workloads at scale. However, while the serverless paradigm offers considerable benefits in terms of scalability and resource management, it also presents unique challenges, including cold start latency, resource allocation, and computational efficiency. This research focuses on a comparative analysis of AI/ML optimization strategies deployed within DevOps pipelines on AWS's serverless architectures. The aim is to identify and evaluate the various optimization strategies available to enhance the performance of AI/ML models, mitigate existing challenges, and improve the efficiency and cost-effectiveness of cloud-based DevOps workflows. This paper reviews optimization techniques such as hyperparameter tuning, model compression, pruning, batch inference, and parallel processing, and their impact on the performance of ML models deployed within AWS Lambda and SageMaker environments. The study involves the empirical evaluation of real-world use cases, providing insights into the trade-offs between model accuracy, resource consumption, and execution time. Key findings suggest that while AWS serverless platforms provide excellent scalability and ease of use, careful management of resources and optimization of workflows is essential to maximize their potential. Furthermore, this paper contributes to the field by proposing recommendations for best practices in optimizing AI/ML workflows in serverless environments, while offering insights into future research directions.