Ai+ Training

These are great times for being a data scientist. At SAS, we see companies of all sizes and industries turning towards model-driven business models as this provides competitive advantages. But often organisational structures are not keeping pace. Friction arises where production IT meets the analytical space and deploying an analytical model becomes a challenge, causing severe delays and high costs. In the area of software development these issues are well known and DevOps has proven to be a good response. 

Enter ModelOps, the adaption of DevOps to analytics. This session demonstrates how SAS can be used to realize the key idea of ModelOps: to put the developer into the driver's seat by giving more control on the model lifecycle. ModelOps is all about automation and toolchains. Git, Jenkins, Docker, Kubernetes typically are the building blocks of these toolchains. We will demonstrate how SAS Model Manager in SAS Viya can be used to fully orchestrate the model lifecycle in open analytical ecosystems. Using SAS and non-SAS scoring models as an example we will cover registering, deploying and monitoring SAS and non-SAS models in Docker containers on a Kubernetes cluster.