AI as a service

Through microservice encapsulation technology, the cost of data scientists’ model service can be reduced and the accuracy difference between new and old models can be better controlled.

AI as a service

Through microservice encapsulation technology, the cost of data scientists’ model service can be reduced and the accuracy difference between new and old models can be better controlled.

Through microservices packaging technology, we not only significantly reduce the cost of data scientist model service and accelerate AI model Preparedness, but also effectively grasp the accuracy difference between old and new models, which is easy for model adjustment and preparation.

Packaging model components

Model packaging (input/output interface, execution version, dependency library and execution platform)
Cross-platform (Redhat, CentosPowerAI) agent deployment
Incorporate the current release process and comply with release regulations

Model versioning and monitoring mechanism

Model monitoring templates with different monitoring indicators according to different model types
Model alerting mechanism to actively warn of deviations to maintain model accuracy
Model and service input records query to restore model predictions at the time

Visual management interface

Visualize model Dashboard for model-related statistics presentation, such as call counts and hotness.
Swagger UI to visualize the analysis service API specification.
Model runtime environment resources Dashboard.

Headquarters 3F., No. 161, Sec. 5, Nanjing E. Rd., Songshan Dist., Taipei City 105409 , Taiwan (R.O.C.)

TEL : +886 2 7701-5899

AI as a service

AI as a service

AI as a service

Packaging model components

Model versioning and monitoring mechanism

Visual management interface

Solution

News

About Us