5 · Install the SDK & run your first job

Python 3.12 required — your functions are shipped to the workers with cloudpickle, which won't unpickle across versions.

Install

# First run — the tarball also bundles the onboarding notebooks:
curl -sL https://pub-c48a651bbb2f42988602aa11bb9d9267.r2.dev/tarball/gridweave-sdk.tar.gz | tar xz
cd gridweave-sdk && python3.12 -m venv .venv && source .venv/bin/activate
pip install gridweave_sdk-*.whl jupyterlab
jupyter lab onboarding.ipynb

# Or into an existing env (Colab etc.) — always pulls the latest version:
V=$(curl -s https://pub-c48a651bbb2f42988602aa11bb9d9267.r2.dev/tarball/sdk-latest-version.txt)
pip install https://pub-c48a651bbb2f42988602aa11bb9d9267.r2.dev/tarball/gridweave_sdk-${V}-py3-none-any.whl

Authenticate and run

Use the token from signup or onboarding, and your platform URL (https://platform.… after HTTPS, else http://<MASTER_IP>:8100):

import gridweave
gridweave.auth("YOUR_TOKEN", platform_url="https://platform.example.com")
gridweave.resources()              # nodes, vendors, and free VRAM

@gridweave.remote(vram="4GB")      # no vram → CPU; vram → a GPU with ≥4 GB free
def matmul():
    import torch
    x = torch.randn(4096, 4096, device="cuda")
    return (x @ x).mean().item()

matmul.run()

Going further

• Full SDK reference — client_sdk/README.md: fractional/multi-GPU, train(), serve() (vLLM + custom images), gather(), storage.
• onboarding.ipynb (in the tarball) — walks every capability end to end.
• LLM-ready docs — your deployment serves its own docs at https://app.…/docs, with a downloadable llms.txt: paste it into an LLM and ask it to build a notebook for your workload.

---

Next: 6 · Update the cluster — ship new versions without losing state.