nubank/fklearn

What it solves

It addresses the gap between model validation and production deployment in machine learning. It aims to ensure that the models used in production match the models that were validated, making the transition from research to deployment production-ready with minimal extra steps.

How it works

fklearn applies functional programming principles to the machine learning workflow. It provides a framework that emphasizes reproducibility, in-depth analysis of results, and a strict alignment between the validation and production environments.

Who it’s for

Data scientists and machine learning engineers who want to apply functional programming patterns to ensure their ML models are reproducible and production-ready.

Highlights

• Support for multiple model backends including LightGBM, XGBoost, and CatBoost.
• Focus on ensuring production models match validated models.
• Designed for reproducibility and in-depth analysis of model results.