Everyone who knows something about DevOps is familiar with the so-called "DTAP" pipeline (Development > Testing > Acceptance > Production) which can be described as the following:
code on a feature branch -> create a PR -> CI pipeline runs, approvers approve -> merge -> CD pipeline runs (that is storing a feature in feature store)
I am not sure I get your question, you deploy your code somewhere, so every time you change the code on a feature branch -> create a PR -> CI pipeline runs, approvers approve -> merge -> CD pipeline runs
If you do not have the deployment yet, CD pipeline would not be part of the process yet
If we have feature store CI/CD pipeline, how to add in above flow?
code on a feature branch -> create a PR -> CI pipeline runs, approvers approve -> merge -> CD pipeline runs (that is storing a feature in feature store)
What if my intention is not to train model just to improve some code functionality, then how can I skip the deployment?
I am not sure I get your question, you deploy your code somewhere, so every time you change the code on a feature branch -> create a PR -> CI pipeline runs, approvers approve -> merge -> CD pipeline runs
If you do not have the deployment yet, CD pipeline would not be part of the process yet