Appreciate it's just an MVP but I think there's a good niche you can go down. Big Data on AWS is such a pain to set-up (Glue, EMR, RedShift, LakeFormation), with IAM policies and roles a simple data pipeline is around 500 lines of YAML. Would be good if you could add native support for that, so say you have some CSV in S3 you want to convert to parquet, drop null fields, and then make shareable with another AWS account. Would solve a massive problem for me