For batch training i have been utilizing sagemaker though it's a bit expensive then ec2 but it's easy to setup and get started.
Make a docker container and push it to ecr then start the training
and track the metrics using any monitoring tool like wandb
If your use case don't require any custom packages then you can
also utilize HuggingFace DLC it which can ...
You probably should
conduct a missing values analysis to see what is the percentage of missing per column (figure below, from dataprep package)
Decide a threshold according to which you may want to completely drop a column or not (depending on how your analysis or model treats nans as well)
For the columns that are not dropped, you should impute the missing ...