Serverless GPU platform built for AI — deploy Python functions, agents and inference endpoints on autoscaling H100s with one command.