Automating MLOps for Deep Learning
MLOps holds the key to accelerating the development, deployment and management of AI, so that enterprises can derive real business value from their AI initiatives. Deploying and managing deep learning models in production carries its own set of complexities. In this talk, we will discuss real-life examples from customers that have built MLOps pipelines for deep learning use cases. For example, predicting rainfall from CCTV footage to prevent flooding. We’ll finish off with a live mask detection demo, showing how to detect individuals who are or aren’t wearing masks in public areas at scale, to help prevent the spread of COVID-19. Throughout these examples, we will share best practices to effectively build, deploy and manage deep learning pipelines in production. We’ll show how to automate, accelerate and scale the data science practice effectively, how to enable continuous development and delivery (CI/CD) of data and ML intensive applications, how to integrate many types of data (images, video etc.) from different sources, and how to set up the foundation to account for an evolving set of use cases of growing complexity.
Deck: Challenge, solution, customers (white labeled) - H2I, SEAGATE, IHS, deep learning, demo (guy- mask detection)