Get Started

1. Sign up

Sign up for Coiled using GitHub, Google, or by making your own username and password.

Sign up with GitHub

google Sign up with Google

Create a username

2. Install Coiled

Install the Coiled client Python library with pip or conda.

pip install coiled "dask[complete]"
coiled login

This will redirect you to the Coiled website to authenticate your computer. It will also save your token on your computer (see API Tokens).

3. Connect to Your Cloud

Run the following command to grant Coiled permission to run in your cloud account.

Coiled creates the IAM policies and network configuration for your account, asking you for permission at each step.

coiled setup aws

Learn more

Alternatively, configure Coiled with custom network configuration (e.g. pre-existing VPC). This is more common for larger organization than for individuals.

coiled setup aws --manual-final-setup

Learn more

Coiled creates the IAM policies and network configuration for your account, asking you for permission at each step.

coiled setup gcp

Learn more

Alternatively, configure Coiled with custom network configuration (e.g. pre-existing VPC). This is more common for larger organization than for individuals.

coiled setup gcp --manual-final-setup

Learn more

Coiled on Azure is currently in public beta.

Our setup CLI will create the service account and network resources for running Coiled in your Azure subscription.

coiled setup azure

Learn more

We’re also happy to chat if you have any special requirements for Azure permissions, network configuration, or anything else! Please contact support@coiled.io if you have any questions or problems.

4. Enjoy 🎉

Your machine can easily create and manage cloud resources, and you can easily run Python code on those resources.

# Create cluster of twenty machines
import coiled

cluster = coiled.Cluster(
    n_workers=15,
)
client = cluster.get_client()
# Run a simple calculation
import dask

df = dask.datasets.timeseries("2000", "2005", partition_freq="2w").persist()
df.groupby("name").aggregate({"x": "sum", "y": "max"}).compute()

Now you’re ready to go! Next, see Manage Hardware.