Thesis supervisor: Ferenc Friedler
Location of studies (in Hungarian): Széchenyi István University Abbreviation of location of studies: SZE
Description of the research topic:
Machine learning (ML) is a method of data investigation for recognizing different patterns. By mathematical models, machines can be trained to predict future datasets. The model is more accurate if it uses more significant quantities of training data. Machine learning workflow necessitates being scaled efficiently to handle a large amount of data in the cloud.
In the data analysis domain, containerization and cloud-native designs are barely applied. The cloud-native applications are independent and scalable stateless microservices that form highly complex workloads. They have been created for the usage of the cloud's potential.
While the major research approach targets machine learning algorithms, the structure of data pipelines has not been investigated in detail. This study examines the orchestration of machine learning workflows like microservices. The cloud-native architecture can leverage and simplify the machine learning lifecycle operations in the cloud, providing the ability to process the constantly expanding datasets. The machine learning models can be developed and deployed in the cloud without any substantial re-design needs.
This research's primary purpose is to identify the machine learning workflows' most efficient deployment strategies through real-life case studies.