Sumatra

It is a platform that offers self-service capabilities to data teams so they may create real-time feature pipelines from event-based data.

CHALLENGES

Since each product strives to get customers' loyalty by increasing their experience, arises a question of how to achieve those things. With Sumatra, products and services can get a customized solution, which is aimed to help Machine Learning engineers in real-time pipelines over raw event data.

SCOPE OF WORK

When our specialist joined the development process, the backend was already implemented. So from our end, we managed the frontend part, to finish the initial product idea by constructing the internal Console system.

location

Austin, Texas, USA

industry

Booking & Management Solutions

technologies

ReactJS

Apollo

Amazon Web Services

services

Web Application Development

Team Extension

Project Overview

Materializations

This page contains a list of materialization statuses and additional information about each materialization. The materialization comprises of timelines, event types and features. They can be run multiple times in a row or deleted. If the materialization is in progress, it gets automatically updated in the table. This list uses data caching from GraphQl, which allows not to make requests for those pages that the user has already visited, but simply to take them from the cache.

Scowl Editor

This panel displays all available to the user brunches with full modification history. On this page the user can get information about the plan, clone the brunch, and start the materialization. With the utilization of such editor, the whole service related to event-based pipelines become easier to manage.

Timelines Panel

Getting data about timelines is possible on this dashboard. Moreover, it tell the user when everything was last updated, who made the changes and the number of present events. This is like timelines backend by warehoused event data for feature engineering, replay and back-fill.