February 26, 2019
ParallelM Solves Real-World Machine-Learning Deployment Challenge with Kubernetes Autoscaling REST Endpoint

SUNNYVALE, Calif., February 26, 2019 ParallelM, the leader in MLOps, today released a new version of MCenter that includes REST-based serving using Kubernetes to create a no-code, autoscaling infrastructure for model serving supporting the leading modeling frameworks. With this release, data scientists can quickly create robust autoscaling REST services for their machine learning models to better serve real-time applications in the cloud or on-premise.

“After working with advanced customers in the financial services industry, we realized that our clients really needed a REST service using autoscaling technology, like Kubernetes, to meet their business requirements for high-volume and low-latency model serving at scale,” said Sivan Metzger, CEO of ParallelM. “So we went out and built it for them while keeping an elegant, simple interface that’s intuitive and integrates with their existing modeling frameworks.”

The 1.3 release of MCenter specifically addresses the deployment challenges of machine learning for real-time, production applications. Unfortunately for many data scientists, many existing data science tools with REST interfaces were designed for testing of model outputs and not for production applications. This means that while these REST endpoints are easy to set up they cannot perform in real-world environments and will fail when they are needed most. The new REST interface in MCenter is intended for real-time serving models with low latency at high volume as is required by real-world applications. By using this more robust REST endpoint, data scientists can be assured that their models will be available to serve their business applications even under the most punishing real-world conditions.

ParallelM’s MCenter gives data scientists the following benefits:

  • Autoscaling with Kubernetes – This new release increases the scalability and performance of ParallelM MCenter across both batch and real-time use cases by using Kubernetes to provide autoscaling infrastructure. Using this industry standard approach allows loads on the infrastructure to scale up and down as needed to optimize resource utilization and manage costs for pay-as-you-go services. Kubernetes also provides robust failover and ease of infrastructure monitoring and management. So, no matter if companies are just starting with machine learning or are already building advanced, real-time AI applications, their platform for ML in production can scale to meet their needs.
  • No Coding Required – The product is easy to use with drag and drop components that require no coding. This expands the number of people who can build out these high-quality pipelines but also reduces errors and risks from custom coding these applications.
  • New Integrated Components – Out-of-the-box components that are pre-configured for the most common modeling frameworks including Scikit Learn, PMML and H2O. This makes it easier to get started with real-time serving for models built on these frameworks in minutes not months.

About ParallelM

ParallelM is the first, and only company entirely focused on delivering machine learning operationalization (MLOps) at scale. ParallelM’s breakthrough MCenter™ solution is built specifically to power the deployment, management, and governance of machine learning pipelines in production so that companies can scale machine learning across their business applications. ParallelM’s approach is that of a single, unified MLOps solution that embeds best practice processes in technology, enabling all ML stakeholders to unlock the business value of AI. Please visit www.parallelm.com or email us at info@parallelm.com.

ParallelM and MCenter are trademarks of Parallel Machines, Inc. All other trademarks are the property of their respective registered owners. Trademark use is for identification only and does not imply sponsorship, affiliation, or endorsement.

Share This Post:

Get started with a free account!

Try MCenter and See How Much Easier ML In Production Can Be

Start Free Account