AI Red-Teaming Specialist (Remote, $10/hr)

About the role

This short‑term contract position focuses on investigating hard‑to‑find facts and verifying information through credible sources to challenge AI models. You will design and execute research tasks that probe model search, reasoning, and answer generation.

Key responsibilities

• Design challenging research tasks to test AI models and write clear benchmark prompts with primary‑source‑backed answer keys.
• Test AI model outputs, analyze search trajectories, source selection, and identify reasoning failures.
• Investigate hard‑to‑find facts, verify information using credible sources, and document findings.

Required profile

• Strong experience in investigative research, fact‑checking, AI benchmark authoring, red‑team testing, or technical QA.
• Familiarity with AI benchmarks such as GAIA, BrowseComp, SimpleQA, SWE‑bench, Terminal‑Bench, or Humanity’s Last Exam.
• Excellent web‑research and OSINT skills, including navigation of academic archives, local records, and specialized databases.

Required skills

• GAIA benchmark
• BrowseComp benchmark
• SimpleQA benchmark
• SWE‑bench benchmark
• Terminal‑Bench benchmark
• Humanity’s Last Exam benchmark
• Web research
• OSINT