Get started with UbiOps
We made several tutorials to get you up to speed. Happy deploying!
We made several tutorials to get you up to speed. Happy deploying!
Pipelines allow users to define sequences of Deployments, by connecting the output of one deployment to the input of another deployment. This is useful when your application depends on a series of separate data transformations that need to operate in sequence.
Example deployments contain ready-made deployment packages for typical use cases, which illustrate how to deploy your Python or R code in UbiOps. Besides the examples treated in the video’s below, you can find more inspiration in our cookbook: https://ubiops.com/docs/ubiops_cookbook/
Image recognition apps are fairly straightforward to deploy on UbiOps and in this deployment package, you can see an example. It is a model that predicts handwritten digits. It takes a picture of a handwritten digit as input and returns its prediction of what digit it is.
To illustrate the basic working of the deployment.py required by UbiOps we have created a sample deployment that multiplies a given number by 2. You can download the deployment package as a zip (ready to be used) and follow the tutorial.
Prediction models are a typical data science application and they are very straightforward to deploy on UbiOps. In the request method, we call model. prediction to actually make the prediction. This structure works for models created with Tensorflow, ScikitLearn or other standard data science libraries.