[R], [P] Self-Hosted GPU setup for AI Research

My 3070 is increasingly holding me back for R&D, and I've been on the cloud more and more not just for running jobs but for active research. I feel like I'm just burning money on the cloud and it's just not sustainable. I need to invest some $$ and time into building a high quality (although smaller still) server to conduct my research.

I've been struggling to find good detailed resources/communities for this. Most people seem to be content with …

ai research building cloud gpu jobs machinelearning money research running self-hosted setup sustainable