Built for Data Scientists
Make R and Python models immediately accessible via standard REST API requests without recoding from their native language.
Embedding models into production apps is as simple as sending an autogenerated code snippet to your dev team. Embed predictive models in any application capable of making REST API requests.
Real-time or Batch
Make real-time predictions in low-latency applications or use batch mode for bulk offline scoring.
ScienceOps ships with version control and tracking to facilitate organized collaboration. Instate a new model via a hot-switch or roll-back to a previous version with the click of a button.
Ensure that your source code operates properly with automatic unit testing before deployment. Unit test new models without interfering with existing models running in production applications.
Use-case: Lending and Financial Services
Access traffic and summary statistics of models running in production. Pinpoint and resolve model errors with speed and precision.
Track and compare model inputs and predictions over time. Inspect individual model inputs and outputs for easy auditing.
System Health Overview
Monitor the stability of your models and servers in real time. ScienceOps ships with a variety of system and model checks in one centralized view.
In addition to monitoring your system within our software, ScienceOps also ships with a built-in Graphite integration for tracking server side metrics.
ScienceOps runs on a clustered architecture composed of one Master server and any number of Worker servers.
Worker servers host predictive models in-memory and run models in discrete run-time environments. Add servers to scale with demand at any time.
Effortlessly replicate models to accommodate seasonal or permanent increases in throughput.
ScienceOps can be installed in a VPC, on-premise or in a hosted Yhat environment.
With ScienceOps your models and your data stay secure on your servers, behind your firewall, and within your data center.