Published: October 31, 2024
Part 3 in our series of blogs on vetiver for MLOps. Having previously introduced the modelling and deployment steps of the MLOps workflow, we now consider the maintenance of a model in production. The monitoring process involves adding a date column to our data, scoring our model at regular time intervals, and checking for signs of model drift over time as the data evolves.
Published: June 20, 2024
Part 2 of our series of blogs on vetiver for MLOps. In this post, we demonstrate how to deploy a machine learning model to production using Docker, Posit Connect, and SageMaker. Docker allows developers to bundle application code with necessary dependencies, simplifying deployment. We outline the process of creating a Dockerfile with the {vetiver} package and running the model locally. Additionally, we show how to publish the model to Posit Connect and SageMaker for broader accessibility.
Published: June 13, 2024
Part 1 of our series of blogs on vetiver for MLOps. This post introduces MLOps and its integration into the traditional data science workflow, focusing on continuous model deployment and maintenance. It demonstrates automating data importation, creating a model with {tidymodels}, and using {vetiver} to store and deploy the model. The process includes creating an API with {plumber} and deploying it locally. Finally, it verifies the API functionality, setting the stage for future production deployments. description: Part 1 of our series of blogs on vetiver for MLOps. This post introduces MLOps and its integration into the traditional data science workflow, focusing on continuous model deployment and maintenance. It demonstrates automating data importation, creating a model with {tidymodels}, and using {vetiver} to store and deploy the model. The process includes creating an API with {plumber} and deploying it locally. Finally, it verifies the API functionality, setting the stage for future production deployments.
