Sagemaker docs

content/docs/extensions/deployment/sagemaker.md

@@ -4,7 +4,114 @@ Implements MlemEnv, MlemDeployment and DeployState to work with AWS SageMaker
 
 ## Description
 
-**TODO**
+MLEM SageMaker allow you to deploy MLEM models to AWS SageMaker. You can learn
+more about SageMaker
+[here](https://docs.aws.amazon.com/sagemaker/latest/dg/whatis.html).
+
+### Preparing infrastructre
+
+To be able to deploy to SageMaker you need to do some AWS configuration. This is
+not MLEM specific requirements, rather it's needed for any SageMaker
+interaction.
+
+Here is the list:
+
+- AWS User Credentials
+- SageMaker access for this user (policy
+  `arn:aws:iam::aws:policy/AmazonSageMakerFullAccess`)
+- ECR access for this user (policy
+  `arn:aws:iam::aws:policy/AmazonEC2ContainerRegistryFullAccess`)
+- AWS IAM Role with SageMaker access
+- S3 Access
+
+You can configure those manually or use existing ones. You can also use
+[terraform](https://www.terraform.io/) with
+[this template](https://github.com/iterative/mlem/tree/main/mlem/contrib/sagemaker/mlem_sagemaker.tf)
+and
+[helper script](https://github.com/iterative/mlem/tree/main/mlem/contrib/sagemaker/env_setup.py)
+(terraform needs to be installed).
+
+> This script is not part of MLEM public API, so you'll need to run it manually
+> like this
+
+```python
+from mlem.contrib.sagemaker.env_setup import sagemaker_terraform
+
+sagemaker_terraform(export_secret="creds.csv")
+```
+
+It's recommended to use [aws cli](https://aws.amazon.com/cli/) with separate
+profile configured for MLEM. You can also provide credentials with
+[AWS environment variables](https://docs.aws.amazon.com/cli/latest/userguide/cli-configure-envvars.html).
+
+### Configuring and running deployment
+
+[SageMaker Environment](#class-sagemakerenv) declaration can be used to hold
+your SageMaker configuration.
+
+TODO
+
+You can also pre-declare [SageMaker Deployment](#class-sagemakerdeployment)
+itself.
+
+TODO
+
+To run deployment, run
+
+```cli
+$ mlem deployment run ... --model <path>
+```
+
+### What happens internally
+
+Once you run our this sweet `mlem deployment run ...` command, a number of
+things will happen.
+
+1. If you did not specify pre-built image, a new docker image will be built. It
+   will include all model's requirements. This image will be pushed to
+   configured ECR repository.
+2. Model is packaged and uploaded to configured s3 bucket as per
+   [this doc](https://docs.aws.amazon.com/sagemaker/latest/dg/realtime-endpoints-deployment.html#realtime-endpoints-deployment-create-model)
+3. Enpoint Configuration is created as per
+   [this doc](https://docs.aws.amazon.com/sagemaker/latest/dg/realtime-endpoints-deployment.html#realtime-endpoints-deployment-create-endpoint-config)
+4. Model is deployed thus creating a SageMaker Endpoint
+
+After this command exits, however it can take some time on SageMakers side to
+actually run VMs with your model. You can check status with
+
+```cli
+$ mlem deployment status ...
+```
+
+or block until model is ready with
+
+```cli
+$ mlem deployment wait ... -i starting
+```
+
+### Making requests
+
+MLEM SageMaker deployments are fully compatible with SageMaker
+[InvokeEndpoint](https://docs.aws.amazon.com/sagemaker/latest/APIReference/API_runtime_InvokeEndpoint.html)
+API, however it's a lot easier to use
+[MLEM SagemakerClient](#class-sagemakerclient). To obtain one, just call
+`get_client` method on your deployment object.
+
+```python
+from mlem.api import load_meta
+
+service = load_meta("...")
+client = service.get_client()
+```
+
+You can then use this `client` instance to invoke your model as if it is local.
+
+```python
+data = ...  # pd.DataFrame or whatever model.predict accepts
+preds = client.predict(data)
+```
+
+> MLEM do not support batch invocations. We will add support for them soon
 
 ## Requirements