10 Cool Platforms For Automating Machine Learning

Here are 10 machine learning platforms and services that solution providers can take advantage of in the enterprise market.

Automatic Intelligence

It wasn't so long ago that developing custom artificial intelligence was very, very hard.

Then came along several software libraries and frameworks, from PyTorch to Keras to MXNet to TensorFlow, that offered precoded neural networks, regression analysis, and other machine-learning models. AI development became easier—but was still very hard.

In recent years, however, building intelligent solutions has finally become possible for those of us who aren't data scientists thanks to a spate of platforms automating the machine-learning pipeline. With demand for AI customized to specific business processes growing far faster than data scientists can be produced and integrated into the workforce, enterprise adoption of those platforms is booming.

Solution providers with analytics chops are taking advantage of automated platforms brought to market by cloud giants and ISVs to preprocess training data; select, implement, train and evaluate models; and deploy those trained models into production environments.

Google Cloud AutoML

Google's automated machine-learning service, Cloud AutoML, accelerates the process of building custom AI solutions with a combination of open-source tools and proprietary technology Google has developed over the last decade.

Cloud AutoML allows developers with limited data science skills to train and tune selected models on distributed infrastructure, including GPUs when desired.

AutoML supports popular libraries, especially Google's homegrown TensorFlow, and offers partially pretrained components for developing custom solutions, like image classification systems and language interfaces, using smaller data sets.

IBM Watson Machine Learning

Before Watson was a comprehensive platform, it was a cognitive application heralding a new era in AI by adeptly beating human contestants on the game show “Jeopardy!”

IBM's Watson division has since incorporated a suite of development components, among them the Watson Machine Learning service, which enables training and deploying machine-learning models on IBM Cloud.

Watson Machine Learning supports all major ML frameworks and provides access to GPUs and highly distributed infrastructure for training models on large data sets.

For developers that want a no-code approach, IBM Watson Studio offers a graphical environment supporting the full AI life cycle, including building and viewing models and training neural networks.

AWS SageMaker

Amazon Web Services offers a number of machine-learning services to accommodate developers across the spectrum of data science expertise.

Increasingly, AWS partners building custom AI solutions for clients are turning to Amazon SageMaker to automate their machine-learning pipelines.

The fully managed environment for building, training and deploying AI at scale on distributed AWS infrastructure is a popular choice because of its preinstalled and optimized algorithms, one-click training and deployment options, and ability to optimally provision GPU-based infrastructure.

SageMaker supports TensorFlow and Apache MXNet frameworks out-of-the-box, and its features are accessible through a high-level Python API.

Microsoft Azure ML

Microsoft's cloud-based machine- learning platform integrates a variety of tools to facilitate enterprise development of AI solutions.

Azure ML offers features that automate the selection of algorithms and the process of pulling models through development and training phases before deployment on Azure cloud. The platform comes with built-in support for many open-source tools and frameworks, including ONNX, Python, PyTorch, scikit-learn and TensorFlow.

Developers can opt for a no-code approach using the Azure ML drag-and-drop interface.


This startup's open-source AI technology is widely adopted among enterprises looking to draw insight from their data.

H2O AutoML eases the machine-learning workflow with automatic training and tuning of models optimized for GPU acceleration. The platform also helps developers select the best models for their tasks and constraints by maintaining a leaderboard of algorithms.

For partners and customers looking for even more automation, H2O offers Driverless AI, a platform that implements advanced data science methods under the hood for rapid solution development.

Domino Data Lab

Domino Data Lab's open data science platform offers a range of development tools, data sets and access to computing resources that fast-track AI solutions into production.

The Domino Data Science Platform allows developers to run their experimental and production workloads in configurable Docker containers for shared and reusable environments. Trained algorithms can be deployed at scale onto on-premises or cloud-based Kubernetes clusters.

The platform offers one-click access to GPUs; supports H2O, TensorFlow and other frameworks; connects to big data systems including Spark and Hadoop and common databases and cloud-based storage.


Alteryx offers a comprehensive platform used by enterprises across industries for developing predictive analytics solutions.

The Alteryx Analytics platform can pull data from many sources—from data warehouses to cloud applications to spreadsheets—then cleanse and integrate that data for training machine-learning models in a repeatable workflow.

The solution aims to be entirely self-service, and all steps of the development pipeline can be designed entirely on a drag-and-drop worksheet, or using Python, PySpark, R and Tensor Flow.

When deploying machine-learning models into production, Alteryx Analytics generates APIs that can be called from various development environments.

Oracle DataScience.com

Oracle bought DataScience.com last year to beef up its AI portfolio with an enterprise ML automation platform that unifies data science tools, libraries and languages with cloud infrastructure to make it easier for data scientists to collaborate and push their work into production environments.

The startup's platform will soon be integrated with existing Oracle technology to deliver the Oracle Data Science Cloud Service—an end-to-end platform for processing data, training models, and deploying them tightly integrated with Oracle Cloud infrastructure.


DataRobot's machine-learning platform is widely used to automate and accelerate the creation of predictive models in industries like banking, health care and retail.

The startup's platform employs massively parallel processing to train and simultaneously evaluate hundreds of machine-learning algorithms developed through open-source libraries. It then selects the model best fitting customer datasets and prediction targets to deploy into production.

The unique technology aims to make predictive models faster and less expensive to deploy for customers that don’t have access to data scientists.


Seldon has created a machine-learning platform agnostic to frameworks, continuous integration and deployment tools, and cloud infrastructure.

The startup's open-source technology, Seldon Core, enables developers to build and tune machine- learning models using an available toolkit or preferred programming language.

The enterprise version offers a managed package for deploying those models and inference graphs across scalable Kubernetes clusters spanning hybrid and multi-cloud cloud environments.