Feature engineering is the process of using domain knowledge to extract features (characteristics, properties, attributes) from raw data that make machine learning algorithms work.
The 'art' side of data science.
Often considered the most important factor in the success of a traditional ML project.