ModelOp Center

The platform allowed us to evolve our model operations practice independent of all our our model development practices through their abstractions. We started our model operations practice before GPT3 and the GenAI era, it is as relevant now as it was then. The model inventory is not just a static documentation to, it drives our model lifecycles making it indispensable to our model operations practice. The abstractions the company provided allowed us to use all of our existing CI/CD, incident management, development management, scheduling, and other enterprise tooling. The product provided us with many efficiencies. Moving models to production used to take up to a year, now they are down to weeks. It made it possible to do ongoing monitoring of models as prior to modelOp monitoring was a periodic task that was manual and could not be done in reasonable times.

1.Configuration. 2.Overall implementation. 3.Support team

Execute model in any language, with the existing Source Code Management tooling.

Model operations is the least understood aspect of our advanced analytics functions. I wish it was easier to teach people about model operations, it's value, and its independence from model development. The ModelOp software, because we chose to integrate it with all of our existing tooling, took longer than anticipated to implement initially. Likewise to onboard new business units takes longer than anticipated. Model operations is still confused with model processes that integrate model pipelines often called MlOps. Maybe we should have named it something different as they really are quite distinct functions.

1.Deployment. 2.Excessive coding 3.technical skills required during implementation.

Product requires professional services for implementation