I’m going to be in-person at posit::conf 2025 in Atlanta, giving a talk about how fundamental R development practices have made model governance (which I’m defining as all the stuff that happens after initial deployment) better for us over the last few years.
As the talk has come together, it’s turning out higher-level than the abstract I submitted earlier this year, but it’s pretty close.
Abstract (from the Session Catalog):
Building Governable ML Models with R
For a model to provide value in production, it must be fit for purpose and deployable into the MLOps environment. We know that R provides a host of tools and packages for building good models; this talk will demonstrate one way we’ve had success assembling R and R-packaging features to provide a stable foundation for model deployment and governance after production. Those attending this talk will learn how, by centering model development on packages, writing tests as we go, creating intuitive S3 methods, and more, we can build modularized, testable code that separates operations from modeling, making our models easier to deploy, monitor, and update, whether the environment incorporates CI/CD, model registries, schedulers, or anything else.