Double machine learning (DML) is a framework for estimation of lower-dimensional parameter vector in a high-dimensional setup, where the number of regressors is potentially very large. This estimation framework has been implemented in a package DoubleML both in R and Python. Besides presenting an overview of the DML framework, the goal of the thesis is to implement estimators from a sample selection/outcome attrition DML extension into DoubleML package, in Python. Such extension deals with situations where outcomes are selectively unobserved (such as wages for unemployed) for a specific subpopulation and is of immediate practical interest for researchers and analysts.