# SkyPilot Quickstart

[SkyPilot](https://skypilot.readthedocs.io/en/latest/docs/index.html) is an open-source framework for running AI and batch workloads on any infrastructure provider, anywhere. Flexibly provision compute in any cloud with the same common configuration-- no switch statements or analyzing instance types.

Tigris enables this extreme portability by globally distributing your data so it's always close to the compute SkyPilot dynamically provisions. Since Tigris is a no egress fee storage provider, you don't have to worry about additional costs by going multi-cloud.

## How to install SkyPilot[​](#how-to-install-skypilot "Direct link to How to install SkyPilot")

There's a few ways to set up SkyPilot, here's the easiest methods:

### Conda[​](#conda "Direct link to Conda")

If you already use [conda](https://anaconda.org/anaconda/conda), follow the [upstream guide](https://docs.skypilot.co/en/latest/getting-started/installation.html) on how to install SkyPilot. Make sure to install the adaptor for your infrastructure:

```
pip install "skypilot[aws]"

pip install "skypilot[kubernetes]"

# ...
```

Or mix and match:

```
pip install -U "skypilot[kubernetes,aws,gcp]"
```

### How to use Docker to run SkyPilot[​](#how-to-use-docker-to-run-skypilot "Direct link to How to use Docker to run SkyPilot")

If you want to skip the grind and just get started, run this `docker` command to set up SkyPilot in a container:

```
# NOTE: '--platform linux/amd64' is needed for Apple silicon Macs

docker run --platform linux/amd64 \

  -td --rm --name sky \

  -v "$HOME/.sky:/root/.sky:rw" \

  -v "$HOME/.aws:/root/.aws:rw" \

  -v "$HOME/.config/gcloud:/root/.config/gcloud:rw" \

  berkeleyskypilot/skypilot
```

Then open a shell with `docker exec`:

```
docker exec -it sky /bin/bash
```

## How to use Tigris with SkyPilot[​](#how-to-use-tigris-with-skypilot "Direct link to How to use Tigris with SkyPilot")

If you're running a training job or storing model weights, reference these guides:

* [Using model weights in Tigris anywhere with SkyPilot](/docs/model-storage/skypilot/.md)
* [Training with Big Data on SkyPilot](/docs/training/big-data-skypilot/.md)

Otherwise, use [boto3 to interact with Tigris](/docs/sdks/s3/aws-python-sdk/.md):

```
# Create S3 service client

svc = boto3.client(

    's3',

    endpoint_url='https://t3.storage.dev',

    config=Config(s3={'addressing_style': 'virtual'}),

)



# List buckets

response = svc.list_buckets()



for bucket in response['Buckets']:

    print(f'  {bucket["Name"]}')



# List objects

response = svc.list_objects_v2(Bucket='tigris-example')



for obj in response['Contents']:

    print(f'  {obj["Key"]}')



# Upload file

response = svc.upload_file('getting-started.py', 'tigris-example', 'getting-started.py')



# Download file

response = svc.download_file('tigris-example', 'getting-started.py', 'getting-started-2.py')
```

Now you have SkyPilot installed with Tigris. Enjoy building!