Hello Raksha!
You can train models and package them into inference pipelines for any cloud environment that can run the required version of Python (3.8 at the moment) and for which platform the prerequisites of AI SDK are available.
In addition, the Image Classification template found in the AI SDK provides explicit instructions on how to use the notebooks in Azure ML Studio which includes an Azure ML pipeline example for training, packaging and delivering a model.
If you have plans to manage your models, the AI Model Manager currently supports cloud-to-edge delivery from Azure and AWS.
Best regards!