Rubric and verifier design

The operating task

Define task-success criteria and reviewable verifiers for complex model outputs. Specialist contribution is shaped around the quality of the signal and its usefulness to model development.

Softinator's contribution

Define task-success criteria that distinguish acceptable artefacts from superficially plausible outputs. The commissioned output includes rubric, verifier pack and assessment examples.

Acceptance in use

Test discriminative behaviour on correct, incomplete and misleading outputs relevant to the task. The engagement defines the baseline, task scope and evidence used to assess the resulting contribution.