ETL integrated into PyTorch

Hi all, I wanted to check the validity of an idea. I was thinking of having integrated support of ETL in PyTorch. Is this feasible ? Wanted to know if someone has done anything around this and if not, any particular reasons for not doing so ?

I’m not sure if ETL stand for “Extract, Transform, Load”, but if so how would PyTorch come into play combining data from multiple systems into a single database?
Could you explain your idea a bit and correct me, if ETL stand for something else?

What I meant was that would it make sense to support different database bindings for Extract-Transform-Load, instead of having multiple frameworks like Spark, Hadoop etc. ?

I’m still unsure, if this is a PyTorch-specific question.
Are you referring to some Spark/Hadoop-to-PyTorch bindings?