Data Scientists are spending too much time trying to get models into production. ParallelM’s MCenter helps you quickly create ML applications in production, automates orchestration, and guarantees machine learning performance 24/7. With MCenter, you can get back to building more models and putting them to work for your business, while maintaining visibility into model health and performance.
MCenter bridges the gaps in the production machine learning to make deployment easy. With MCenter you can work with your existing data science language or platform like Python, R, Jupyter, H2O and more. Models can be deployed for batch and streaming inference in minutes. Best of all, you don’t have to worry about infrastructure and engines – just provide your model code and MCenter does the rest, like creating and managing Docker containers or deploying on Spark, Tensorflow, PySpark and more.
MCenter provides unique capabilities for monitoring ML applications in production. With MCenter, you have complete tracking of over 150 model and production metrics, without writing any monitoring scripts – it’s all automatic. Build deep custom instrumentation into your production models with the MLOps API. Best of all, MCenter continuously monitors your models and alerts you when you need to pay attention.
MCenter delivers patented ML Health capabilities to help you optimize ML performance in production. Detect production data deviations/drift and data asymmetry between training and production data sets that can lead to poor prediction performance with patented ML Health algorithms. Explain models to business stakeholders with built-in explainability, including feature importance charts and more. Optimize model performance in production with champion/challenger pipelines, and control/canary pipelines with A/B testing.
The MCenter workspace ensures you have visibility into your ML applications in production including intuitive dashboard views for ML Health indicators, KPIs, and advanced visualization and diagnostic tools for model monitoring.
The MCenter server orchestrates ML applications and pipelines via the MCenter agents. It executes policies, manages configuration, and sends data to the MCenter console. The MCenter server enables automation of all the key tasks related to deployment and management of ML.
The MCenter agents trigger analytics engines and manage local ML pipelines. They provide visibility into the activity of the pipeline and send alerts, events, and statistics to the MCenter server.
The MCenter Developer Connectors allow you to easily bring your models into MCenter from leading data science platforms such as Jupyter, Apache Zeppelin, DataRobot, H20.ai, Cloudera Data Science Workbench (CDSW), IBM DSX and more.
Write code in any language (R, Python, Java, Scala, etc.) Run code either on any analytic engine (Spark, TensorFlow, Pytorch, etc.) or natively in a Docker container (R and Python).
MCenter can be deployed in any cloud, on-premise, hybrid or airgap scenarios.