Hi! I’m building InferWeave, an open-source Python SDK that makes deploying AI models on cloud GPUs easier to manage. Instead of wiring every application to a different provider API, you can use one async interface to launch a model, connect to its inference endpoint, check on a deployment, recover it later, and shut it down when you’re done.
The project brings together provider integrations such as Modal, Lightning AI, RunPod, and Vast.ai, along with runtime configuration, health checks, and deployment lifecycle tools. My goal is to make it simpler to get a model running on GPU infrastructure and keep track of it without hiding the details that matter.
InferWeave is still evolving, and I’ll share progress and the challenges I run into here. You can explore the code on GitHub: https://github.com/AmaLS367/InferWeave