A Microservices-Based Approach for Scalable Deployment of Machine Learning Models on a Cloud-Based Platform

Authors

Rahul Roy Devarakonda

Abstract

The need for scalable, effective, and adaptable deployment methodologies has increased dramatically in tandem with the quick growth of machine learning applications. Adaptability to changing workloads, resource optimization, and scalability are common issues with traditional monolithic systems. To facilitate modularity, scalability, and ease of integration, this study examines a microservices-based approach for deploying machine learning models on a cloud-based platform. The proposed design leverages distributed computing, orchestration, and containerization concepts to achieve fault tolerance and optimize resource utilization. The microservices strategy reduces deployment complexity and enhances system performance by decoupling various components, including data preprocessing, model inference, and result aggregation. Furthermore, cloud-native tools are integrated to optimize computational costs, simplify model scaling, and expedite workflow execution. This study also highlights the difficulties with load balancing, API connectivity, and model interoperability, along with potential solutions using dynamic orchestration frameworks. The findings show that, compared to conventional monolithic techniques, a microservices-based deployment significantly enhances response speed, fault tolerance, and resource efficiency. The results provide a scalable and effective framework for practical applications, which advances cloud-based AI deployments.

Keywords

Microservices Architecture, Scalable Machine Learning Deployment, Cloud Computing, Fault Tolerance in ML Deployment, Auto-Scaling ML Models, MLOps and CI/CD Pipelines

A Microservices-Based Approach for Scalable Deployment of Machine Learning Models on a Cloud-Based Platform. Rahul Roy Devarakonda. 2017. IJIRCT, Volume 3, Issue 1. Pages 1-7. https://www.ijirct.org/viewPaper.php?paperId=2503072