Practitioner-authored
Examination items are grounded in real deployment decisions — drawn from the failure modes, tradeoffs, and edge cases that arise in production AI environments. Items are not sourced from academic literature or vendor documentation.
Framework-anchored
Every item maps to a specific PSF domain and competency level. The exam blueprint (published at /certify/blueprint) shows the weighting of each domain. Passing the exam demonstrates coverage across all eight domains, not just the ones a candidate finds comfortable.
Scenario-based
Items present realistic production scenarios — an AI system behaving unexpectedly, a data governance decision, an architecture choice — rather than testing vocabulary recall. The correct answer requires applying PSF principles, not memorising definitions.
Regularly reviewed
The item bank is reviewed annually. Items with poor discrimination (where high performers and low performers choose answers at similar rates) are flagged for revision or removal. Items are also reviewed when the PSF is updated to ensure continued alignment.