Certified AI Data Governance Professional

What you will be able to do

• Set up ownership and stewardship for data used by AI
• Define and monitor data quality rules suited to model training
• Document lineage from source to model output
• Classify data and control its use for training and inference
• Manage training data, labels and datasets as governed assets
• Use AI itself to speed up governance tasks

Modules

1. 1Governance foundations for AI

   The governance disciplines that already exist, and what AI adds to each.

   • Governance frameworks and their knowledge areas
   • Policies, standards and procedures
   • Owners, stewards and custodians
   • Governance bodies and decision rights

   Exercise. Assign an owner and steward to five datasets that feed a model and write their responsibilities.

2. 2Data quality for training and inference

   Quality measured against the purpose: what is good enough for a report may not be good enough to train on.

   • Fitness-for-purpose quality dimensions
   • Profiling and rule definition
   • Representativeness and class balance
   • Quality monitoring in production

   Exercise. Write ten quality rules for a customer dataset that will train a churn model.

3. 3Metadata, catalogs and lineage

   Being able to say where data came from and what was done to it, all the way to a model's decision.

   • Business, technical and operational metadata
   • Data catalogs and glossaries
   • End-to-end lineage including feature pipelines
   • Datasheets for datasets and model cards

   Exercise. Draw the lineage from source systems to one model output and note each transformation.

4. 4Classification, privacy and permitted use

   Knowing what may be used for which purpose, and enforcing it.

   • Sensitivity classification schemes
   • Personal data, consent and purpose limitation
   • De-identification and its limits
   • Access controls for training environments

   Exercise. Classify the fields in a dataset and mark which may be used for training without further approval.

5. 5Training data as a governed asset

   Datasets, labels and versions recorded so that a model can be reproduced, audited and retrained.

   • Dataset versioning
   • Labeling standards and inter-rater agreement
   • Synthetic data and its governance
   • Retention of training data

   Exercise. Write a labeling guideline for one classification task, including how disagreements are resolved.

6. 6AI for governance

   Using AI to discover sensitive data, suggest glossary terms, detect quality issues and draft documentation.

   • Automated sensitive-data discovery
   • Suggested metadata and glossary mapping
   • Anomaly detection for quality
   • Keeping stewards accountable for AI suggestions

   Exercise. Pick one governance task you do by hand and design how AI would assist it, with the steward's check.