Sanatio - Fluid AI

PythonLibraryCICD

Sanatio(Latin) OR validation(English)

Sanatio is a Python package targeted at increasing a developer's productivity during the machine learning model life cycle. Its primary goal is to relieve a developer's repetitive work on the validation and development sides so that he or she can devote more time to model enhancement research.

I was the DRI for the entire library development. Some of the key features I developed in Sanatio are:

  • Model Validation Report
  • Enact module that implements the recommendation from validation report
  • Automl pipeline integration
  • Code generation in jupyter notebooks

Along with the feature development, I also implemented the supplemental developments that include :

  • The library is built with test cases to be robust
  • The current version of library has a coverage of 81%
  • A server authentication mechanism to authenticate user access token
  • Gitbook user friendly documentation and technical documentation in Wiki
  • Built the CICD stages to push versions of library and also to verify the test cases with sonarqube quality checking

The library is available at pypi at https://pypi.org/project/fluidai-sanatio/

The gitbook documentation is available at https://fluid-ai.gitbook.io/fluid-ai-sanatio