I built a tool that runs a 2-minute benchmark on your machine and predicts how long your AI job will take on T4/V100/A100 before you spend anything. Tested on

I built a tool that runs a 2-minute benchmark on your machine and predicts how long your AI job will take on T4/V100/A100 before you spend anything. Tested on 3 real machines. CPU↔CPU correlation r=0.9969. Would love feedback from anyone running ML workloads.