Artificial Intelligence/Machine Learning Ops

Trend 9: Emergence of enterprise-scale AI model management

As AI initiatives graduate from proof of concept to enterprise deployment, organizations face challenges with the engineering complexity of model deployment, the ability to scale infrastructure efficiently and the lack of AI model visibility and governance. A few large organizations have started investing in the construction of an enterprise-scale AI model management framework. They are developing a repository of AI artifacts such as models, pipelines, features and datasets and managing a complete lifecycle of AI models that range from tuning and training to deployment and monitoring. This approach enables businesses to execute AI in a poly-compute manner and ensure AI governance through the detection of drift and bias, and the explanation and reproduction of AI prediction. While cloud providers offer fully managed model lifecycle management, the open-source community also contributes significantly to create frameworks such as Kubeflow on Kubernetes’ containerized platform.

Infosys partnered with a major U.S.- based finance organization to create a centralized AI platform for model training, management and deployment.

Artificial Intelligence/Machine Learning Ops

Trend 10: Integrated development environment for data science projects

Data scientists need different development environments, depending on the nature of data and their AI framework. Traditionally, data scientists have used various development tools to choose their desktop to develop ML models, but deploying and integrating them with their AI ecosystem has been a challenge. Organizations are now setting up environments where a data scientist can perform various tasks like data acquisition, profiling and pipeline development; obtain enterprise assets like datastores, models, Git and more; access different packages/frameworks; collaborate with other team members; integrate with a model management framework or deploy the model at scale – all within the same development environment and context of the AI project.

A U.S.-based telecommunications company democratized AI by building an AI platform for their data scientists, data engineers and business analysts by partnering with Infosys. The new platform allows them to set up challenges, compete and collaborate across their organization, as well as integrate with AutoML tools.

Subscribe

To keep yourself updated on the latest technology and industry trends subscribe to the Infosys Knowledge Institute’s publications

Infosys TechCompass