Input
What you provide
- Source examples, task instructions, and the rubric.
- Modality and usage rights.
- Required expertise and the quality threshold to clear.
- Escalation rules and the delivery format.
AuraOne / Human Data / Annotation
Annotation
Create, rank, and correct training examples with the judgment attached.
Every dataset starts from an explicit task and rubric. Uncertain work goes to review. Quality, rights, and delivery records leave with it.
record annotation-01Training-ready dataset
- Task definition
- Annotation type and modality, with the rubric and examples
- Reviewer readiness
- Required expertise and calibration result
- Quality result
- What was sampled, what was disputed, and how it resolved
- Delivery manifest
- Schema and file count. Rights, exclusions, and approval.
Quality claims apply to the approved sample method and task definition.
Input, work, and output
Work
What happens
- Calibrate reviewers, then label or rank the examples.
- Run quality checks and adjudicate the disagreements.
- Retain rubric versions, reviewer actions, and exception decisions.
Output
What you receive
- Accepted examples in the requested schema and file format.
- A quality summary and the adjudication log.
- The rights record and the delivery manifest.
Decision flow
Move from intake to review and handoff with clear owners at every step.
01
Define and calibrate
Confirm the instructions and the rubric. Confirm reviewers are ready before production.
Task version and calibration result
02
Create the examples
Label it, rank it, or correct it. Always against the approved task.
Item-level action and reviewer identity
03
Review and adjudicate
Check quality, resolve disagreements, and retain exception decisions.
Quality sample and adjudication log
04
Approve delivery
Validate the schema and the rights. Check thresholds and manifest before handoff.
Delivery approval and signed manifest
What you receive
Annotation deliverables
- Task definition
- Annotation type and modality, with the rubric and examplesVersioned instruction set
- Reviewer readiness
- Required expertise and calibration resultWorkforce qualification record
- Quality result
- What was sampled, what was disputed, and how it resolvedQuality workflow
- Delivery manifest
- Schema and file count. Rights, exclusions, and approval.Export and release gate
What the workflow supports
| Outcome | Work | What you receive | Program fit |
|---|---|---|---|
| Training-ready dataset | Expert annotation, review, and adjudication | The rubric, every item action, and the manifest | Quality claims apply to the approved sample method and task definition. |