What is an application

This document describes the function of applications in the Seaplane platform and how to use them.

The application, sometimes called pipeline forms the heart of any data science or machine learning pipeline deployed on Seaplane. The app provides the entry point to the application and contains the directed acyclic graph (DAG) that indicates the flow of data.

Apps are created with the @app decorator. Once configured and deployed the application sets up all the required infrastructure such as API gateways, streams between components, compute containers and much more.

Inside the @app decorator, each application should at least have an ID and an entry point or trigger. The entry point or trigger to your application can be powered through HTTP, Kafka streams or a Cron job. You can learn more about entry points to your application here.

The example below shows an HTTP POST request-enabled application.

from seaplane import app

@app(id='my-app-id', path='/my-api-endpoint')
def my_application(body):
    # your task logic here

The body object between the function brackets contains the data object submitted to the application through the entry point.

The flow of the data is indicated by the DAG defined inside the function definition below the @app decorator. It wires together different tasks to get the desired result.

info

Tasks are the core building blocks inside Seaplane applications. You can learn more about them here.

The example below shows a simple linear flow through tasks a, b and c. However, much more intricate flows are possible to create complex applications easily. You can learn more about them here.

The code below shows how to implement this flow in an application. It assumes tasks a, b and c are available to the application and defined elsewhere.

from seaplane import app
from ... import a, b, c

@app(id='my-app-id',path='/my-api-endpoint')
def my_application(body):
    # defining the DAG
    output_a = a(body)
    output_b = b(output_a)
    output_c = c(output_b)