RAG Retrieval Error Analysis Specialist
Quick facts
- Pay
- $12–$18/task
- Type
- Contract
- Location
- Remote
- Platform
- Globik AI — 15 open jobs
- Payout
- Not published by Globik AI
- Apply
- Free to apply — never pay to get AI training work
Globik AI is seeking RAG Retrieval Error Analysis Specialists to identify and categorize the reasons retrieval systems fail to return useful information.
You will analyze retrieval failures such as irrelevant results, missing evidence, duplicate passages, poor context selection, and incorrect source matching, then record structured failure categories.
Responsibilities
- Review failed RAG retrieval examples.
- Identify whether the failure is caused by irrelevance, missing information, duplication, poor matching, or another project-defined issue.
- Compare retrieved content with the user's query and expected evidence.
- Categorize retrieval failure types consistently.
- Identify recurring failure patterns and edge cases.
- Document findings for evaluation and engineering teams.
- Maintain accurate error labels across large datasets.
- Participate in calibration and quality reviews.
Pay
$12–$18/task
Apply by 2026-11-12.