Data/ML/AI
Data Labeling
Also written as Data Annotation, Ground Truth, Human-in-the-Loop
Paying people to mark up examples — this email is spam, this box is a pedestrian — so a model has something correct to learn from. Usually the slowest and most expensive part of a project.
Think of it like
Writing the answer key before you can set the exam.
Junior or senior?
Anyone who has really done it talks about labellers disagreeing with each other.
Senior sounds like
Measured that agreement and can say what they did when it was poor.
Ask them
“How often did two labellers disagree, and what did you do about it?”