Data Quality, Labeling, and Bias

Quality gates with Glue Data Quality and DataBrew, labeling pipelines, train/validation/test splits and the leakage trap, class imbalance, bias mitigation, and GenAI data integrity.

Content current as of 2026-10.

Lessons

1. Quality gates: Glue Data Quality and DataBrew profiles
2. Labeling and annotation pipelines
3. Splits, shuffles, and the leakage trap
4. Class imbalance and bias mitigation
5. GenAI data integrity: prompt–response pairs and safety screens

Sources

• AWS Glue Data Quality
• Amazon SageMaker Ground Truth
• SageMaker Clarify — measure pre-training bias