Skip to content
Machine Learning Serving cluster
Branch: master
Clone or download
HydroRobot HydroRobot
HydroRobot and HydroRobot Publishing new version 2.0.4
Latest commit 60666f9 Apr 19, 2019

Hydrosphere Serving

Join the chat at

Installation guide:
Getting started: is the first open source platform for Data Science Management automation. It delivers reliability, scalability and observability for machine learning and AI applications in production. automates deployment and serving ML models, monitoring and profiling of production traffic, monitoring of models performance, data subsampling and model retraining.

The platform makes more Data Science and less data plumbing and tinkering happen.

Hydrosphere Serving enables you to get your models up and running in an instant, on just about any infrastructure and using any of the available machine learning toolkits. It lets you monitor your models’ performance, analyse their inputs (for example, determine whether there is recently an increased number of outliers or not), observe models’ inference on given data and so on.


  • Serverless user experience in your data center or public cloud.
  • Plumbing. Automatic generation of Protobuf contracts for REST, gRPC and Streaming Kafka API from the model metadata
  • Safe experiments and models warm up on shadowed or canary traffic
  • Unified across ML frameworks
  • Automatic data profiling and statistical check of the model quality
  • Immutable model versioning
  • Agnostic to training pipeline and notebook environment
  • Multi-framework Pipelines (e.g. Scikit-learn -> Spark ML -> TensorFlow pipeline)
  • Out of the box tuned and optimized Serving Runtimes
  • Models optimization for Serving

Check UI - explore, deploy and test models

Hydrosphere Serving Deploy Models

Hydrosphere Serving Create an AI application

Hydrosphere Serving Test ML Models

Related repositories

You can’t perform that action at this time.