RAG Answer Grounding Specialist

Globik AI; United States; À distance; USD 21 000 - 28 000; Plein temps

Résumé du poste

Globik AI is seeking RAG Answer Grounding Specialists to evaluate whether AI-generated answers are accurately supported by the documents or knowledge sources provided to the model. You will review AI answers alongside source material and identify unsupported claims, hallucinations, contradictions, and statements that go beyond the available evidence.

Responsibilities include reviewing answers with sources, checking evidence, and documenting grounding errors with evidence.

Qualifications

• Experience with RAG search information retrieval, research content QA or AI evaluation is a plus.
• Comfortable following detailed evaluation rubrics.

Responsabilités

• Review AI-generated answers alongside source documents.
• Check whether factual claims are supported by available sources.
• Identify unsupported statements and hallucinations.
• Identify contradictions between the answer and source material.
• Determine whether the response accurately represents source information.
• Check whether important qualifications or limitations were omitted.
• Compare answers with reference answers when available.
• Categorize grounding errors.
• Document unsupported or contradictory claims with evidence.
• Provide structured feedback for improvement.

Connaissances

RAG evaluation

Information retrieval

Quality assurance

Detail-oriented

Description du poste

Globik AI is seeking RAG Answer Grounding Specialists to evaluate whether AI-generated answers are accurately supported by the documents or knowledge sources provided to the model.

You will review AI answers alongside supporting source material and identify unsupported claims, hallucinations, contradictions, and statements that go beyond the available evidence.

Responsibilities

• Review AI-generated answers alongside source documents.
• Check whether factual claims are supported by available sources.
• Identify unsupported statements and hallucinations.
• Identify contradictions between the answer and source material.
• Determine whether the response accurately represents source information.
• Check whether important qualifications or limitations were omitted.
• Compare answers with reference answers when available.
• Categorize grounding errors.
• Document unsupported or contradictory claims with evidence.
• Provide structured feedback for improvement.

Qualifications

• Experience with RAG search information retrieval research content QA or AI evaluation is a plus.
• Comfortable following detailed evaluation rubrics

Pay

$15–$20/task