Simplify deployment of flows on prefect by using flow-builder.

Given most folks on a team are not prefect deployment experts, it is useful to have a simple way to deploy flows.

This library makes use of the builder pattern to simplify the deployment of flows. It also provides a sample implementation of the deployer for deploying flows to an EKS kubernetes cluster. This makes deployment of flows as simple as the interface provided by the deployer avoid a relatively steep learning curve of prefect deployment concepts.

The deployer also provides a simple CLI tool to deploy flows.

Using pip:

pip install perfect_deployer

Create your custom deployer by implementing the FlowBuilderInterface class, or use one of the implementations provided by the library.

For example to deploy flows:

• to an EKS cluster
• persist results to s3
• use a dask-kubernetes task runner
• infer deployment metadata from the flow

you would write a snippet like this

from perfect_deployer.implementations.infra import eks as kubernetes
from perfect_deployer.implementations.task_runner import dask_k8s as dask
from perfect_deployer.implementations.metadata import deployment_metadata_from_flow
from perfect_deployer.implementations.filesystem import S3Filesystem

@kubernetes(
    cpu=0.8,
    memory=1.5,
    image="my-image",
    service_account="my-service-account",
)
@dask(
    num_workers=5
)
@deployment_metadata_from_flow(environment="dev")
@flow(
    result_storage=S3Filesystem(
        type="s3",
        bucket="my-bucket",
        path="my-path",
    )
)
def simple_flow(x: int, y: int) -> int:
    """Add two numbers x and y."""
    return x + y

deployment = simple_flow.build_deployment()
deployment.apply()