On the Leeway of MLOPs, you have plenty of open-source tools available to build your pipeline. I have picked up mlflow to track & maintain my NLP lifecycle while on the cloud counterpart, for model registry & deployment I have chosen azure ( because why not when you get most of the services in the free tier :P)
Here, We're running an mlflow experiment in which we're building a NER model which is being tracked by MLflow on azure cloud.
To understand the process, I've written up the blog, do check out: https://impyadav.medium.com/mlops-deploy-nlp-model-straightforward-using-mlflow-azure-2df9293d5378