AGL has rebuilt its analytics capabilities for running a central platform that is powered by the Azure service instead of running their projects from standalone virtual machines or laptops. According to a new case study, the new platform has gone live in the mid-2020 and also it forms the technological foundation for the AGL’s data and analytics center of excellence which was created last year.
Joey Chua who is the senior manager of machine learning engineering said that they needed a platform that will allow then to manage the models, codes, and data as a whole rather than as individual bits and pieces. Also, it was their motivation for reaching out to Microsoft to see whether they had a tool that could build on.
AGL said in June that this year they would move almost all their computing to the cloud by 2022 with most of the system destined to run on Azure, it is also said that data and AI services were among the key ones it would consume. But the company had declined to elaborate on its uses cases on time. A package of the Azure services had been mainly used for creating a single platform that combines and centralizes the analytics tools, data science resource, and also machine learning which are widely used by AGL.
AGL also said that it is piloted the architecture before they went for progressing the production deployment of the platform during the mid-2020. It mainly provides integration of different Azure services such as Azure Kubernetes Service and Azure Datatricks by giving AGL a consistent, preconfigured environment by including all artifacts code controlled, managed, and also documented.
Azure Datatricks also provides big data and analytics and powerful data engineering tools in a standardized workspace which can help the AGL’s machine learning teams to fully capitalize on the model building and also for training capabilities of the Azure machine learning. The production environment also tales the advantages of Azure Key vault and also application insight, Azure monitor log features in an Azure monitor, Azure storage, etc.
By using all these services together AGL has a highly secure and efficient way to train, deploy, and manage thousands of models in a parallel way. Also, they believe that this one platform will meet the needs of all different applications.