Optimising Cloud Infrastructure to Support Large-Scale Machine Learning Workloads

Poppad, Rahul (2025) Optimising Cloud Infrastructure to Support Large-Scale Machine Learning Workloads. Masters thesis, Dublin, National College of Ireland.

Preview	PDF (Master of Science) Download (1MB) | Preview
Preview	PDF (Configuration Manual) Download (255kB) | Preview

Abstract

This research discusses the deployment of cloud-based infrastructure with optimal configuration to handle large-scale machine learning (ML) workloads efficiently and scalable economically using Amazon Web Services (AWS). The main focus is leveraging the use of AWS services like EC2, S3 and SageMaker in making the deployment, training and inference of machine learning models easier. It focuses on the important issues encountered in the cloud environment during handling ml workloads, like resource allocation, data storage and model deployment. It assesses Amazon EC2 for scalable compute resources, Amazon S3 for data storage, and Amazon SageMaker for automation of model training, hyperparameter tuning, and deployment. Amazon Web Services (AWS) is an integral part of many businesses and an incalculable host in creating a cloud environment where any business can improve different services. This paper presents a novel cloud infrastructure solution to palette various applications of AWS like mobile pricing, air quality prediction, cardiac anomaly detection and industrial defect classification of an optimized AWS architecture suitable for real-time inference. Performance benchmarks highlight the ability of the proposed infrastructure to efficiently accommodate big ML models over a wide range of working conditions with optimized cost-resource utilization. Cloud-Based Solutions for Scalability and Efficiency in Machine Learning Systems: AI and Data Analytics Machine learning systems are growing rapidly, but if the scale of their implementation is not properly managed, their effectiveness will be reduced. Hybrid cloud environment and edge computing are also opportunities for future work to further gain on performance and cost.

Item Type:	Thesis (Masters)
Supervisors:	Name Email Heeney, Sean UNSPECIFIED
Uncontrolled Keywords:	Large-scale machine learning; Data storage; Cloud environment; Air quality prediction; Hybrid cloud environment
Subjects:	Q Science > QA Mathematics > Electronic computers. Computer science T Technology > T Technology (General) > Information Technology > Electronic computers. Computer science T Technology > T Technology (General) > Information Technology > Cloud computing Q Science > Q Science (General) > Self-organizing systems. Conscious automata > Machine learning
Divisions:	School of Computing > Master of Science in Cloud Computing
Depositing User:	Ciara O'Brien
Date Deposited:	25 Nov 2025 17:47
Last Modified:	25 Nov 2025 17:47
URI:	https://norma.ncirl.ie/id/eprint/8958

Actions (login required)

View Item