Introduction

Predicate allows testing models and checking data quality by calculating metrics according to regulations and generating reports.

The main backend service is a FastAPI server that accepts requests at the address described in the service's ingress. Some tasks are performed asynchronously using Celery. Data is stored in a Postgres database. For each project, a DAG Airflow is generated, which allows monitoring data, metrics, and models according to regulations.

User interaction with the application is carried out through a web interface. Access to functionality is granted after authorization.

Installation

The initial installation of the software is performed by the development and implementation team of OOO "Data Sapiens" on a server provided by the customer.

Update

Checking, downloading, and installing software updates requires an Internet connection and is performed manually on the server by the software administrator. To update a software component, it is necessary to remove it and reinstall it, after which Kolmogorov.ai Predicate should be restarted.

Reinstallation

Any available version of Kolmogorov.ai Predicate can be downloaded and installed using containers provided by the developer. Before reinstallation, the existing software may be removed.