/pipeline

PipelineIO: End-to-End ML and AI Platform for Real-time Spark and Tensorflow Data Pipelines

Primary LanguageJupyter NotebookApache License 2.0Apache-2.0

PipelineIO

Built with Love and Scale by ex-Netflix, Databricks, and Tensorflow Engineers

Tensorflow Databricks Netflix

Reference Architecture

Reference Architecture

100% Open Source

Jupyter ZeppelinZeppelin Spark Tensorflow

Completely Kubernetes and Docker Based

PipelineIO Kubernetes Kubernetes Docker

PMML + Native C++/Java Code Generation for Optimized CPU/GPU Performance

Generate and Optimize Spark ML Model Nvidia GPU Intel CPU

Deploy directly from Jupyter Notebook to Hybrid-Cloud Microservices

Deploy Spark ML Model to Production

NetflixOSS-based Microservices for High Scalability and Availability

NetflixOSS-based Microservices

Continuous Train/Deploy Spark ML and Tensorflow AI Pipelines to Production

Kafka Kinesis KINESIS

Environment Setup

Pipeline Mini (Local Laptop)

  • Small Kubernetes cluster for your laptop

Kubernetes Docker

Pipeline AWS

  • Kubernetes cluster on any cloud provider or on-premise

AWS

Pipeline Google

  • Kubernetes cluster across many cloud providers and/or on-premise

Google Cloud Platform

24x7 Global Support for Everyone!

Additional Resources

Training Repo and Workshop Schedule

Workshop Architecture

Workshop Tools

Global Advanced Spark and Tensorflow Meetup

Meetup

Slideshare Slides

Slideshare

YouTube Videos

YouTube

Docker Images

Docker