Modal

This lets Claude deploy and run Python code on Modal's serverless platform, with first-class support for GPU workloads. You can spin up containers with H100s or A100s on demand, serve ML models as web endpoints, schedule batch jobs, and manage persistent storage for model weights. Everything is defined in Python code instead of YAML or Dockerfiles, which keeps the setup simple. The skill covers the full workflow from authentication through deployment, including volume mounts, secret management, and autoscaling configuration. Modal offers sub-second cold starts and scales from zero to thousands of containers automatically. If you're deploying AI models or running GPU-accelerated workloads without managing infrastructure, this is the straightforward path.