Deploying SageMaker models using CI/CD

in addition to _takanash

i can give you following link to crate your pipeline

https://www.edlitera.com/blog/posts/aws-sagemaker-ci-cd-pipelines

in summary

• You'll need to create a SageMaker Studio instance, enable SageMaker project templates, and create a new user that uses the same role as the Studio instance. Note that if you didn't set up your Studio to allow project templates in the beginning, you won't be able to use these templates​.
• SageMaker Studio provides three different templates: "MLOps template for model deployment", "MLOps template for model building and training", and "MLOps template for model building, training, and deployment". You should select the template that includes model building, model training, and model deployment, which allows you to look at and analyze an example of a complete end-to-end workflow​1​.
• Once you create a new project, SageMaker creates two repositories for model building and model deployment. You should clone both repositories, which contain all the artifacts needed to create a full end-to-end workflow​1​.
• You write individual steps, define those steps with parameters, and afterward connect the steps in a specified order to create a pipeline​1​.
• This can be separated into two parts. First, you need to define the processor and afterward the processing step itself​1​.
• First, you need to define the model you are going to use alongside the parameters of that model. Afterward, you need to define the step itself​1​.
• Once the model is trained, you should register it. This is done in a similar way to how you define the previous steps. You need to define the model you are going to use, alongside the parameters of that model. Then you need to define the step itself. Unfortunately, the specifics of this step were not visible in the source I found.