AI Safety Response Evaluator

Globik AI; United States; Remoto; USD 22.000 - 28.000; Tempo pieno

Descrizione del lavoro

Globik AI is seeking AI Safety Response Evaluation Specialists to review model outputs against defined safety policies and determine whether responses are safe, helpful, and aligned with guidelines.

You will score responses, flag violations, provide rationale for each judgment, and escalate high-risk cases according to project protocol. This role demands strong written communication and attention to detail as you document decisions and participate in ongoing refinement of the evaluation rubric.

Mansioni

• Review model responses against a defined safety and helpfulness rubric.
• Score safety, harmlessness, and helpfulness dimensions.
• Flag policy violations and categorize violation types.
• Distinguish appropriate refusals from over-refusals or under-refusals.
• Provide written rationale supporting each evaluation.
• Escalate ambiguous or high-risk cases per project protocol.
• Maintain consistent scoring standards across large evaluation batches.

Descrizione del lavoro

Globik AI is seeking AI Safety Response Evaluation Specialists to review model outputs against defined safety policies and determine whether responses are safe, helpful, and aligned with guidelines.

You will score responses, flag violations, provide rationale for each judgment, and escalate high-risk cases according to project protocol. This role demands strong written communication and attention to detail as you document decisions and participate in ongoing refinement of the evaluation rubric.