LLM Trainer - Agent Function Call
Design and craft multi-turn conversational datasets to improve AI agent reasoning, function calling accuracy, and real-world interaction capabilities.
$70–$90/hour
Evaluate the quality, fidelity, and completeness of vulnerability-reproduction and remediation tasks used to train and evaluate a frontier AI lab's models. You'll assess whether CVE reproductions are faithful, fixes are sound, verification logic is rigorous, and Docker-based lab environments accurately recreate exploitable conditions — and provide clear, rubric-based written feedback.
Basic Qualifications
• 3+ years of hands-on experience in application security, penetration testing, or vulnerability research
• Strong understanding of CVE vulnerability taxonomy and severity frameworks (CVSS, CWE, CAPEC)
• Demonstrated expertise in secure coding and remediation across common vulnerability classes (SQL injection, command injection, buffer overflow, deserialization, SSRF, misconfigurations, privilege escalation)
• Experience designing or evaluating two-part verification logic (functionality tests + vulnerability tests)
• Proficiency with Docker and Docker Compose for multi-container vulnerability reproduction environments
Preferred Qualifications
• OSCP, GPEN, GWAPT, or equivalent offensive-security certification
• Experience with CVE disclosure, responsible vulnerability reporting, or maintaining exploit proof-of-concept code
• Background in DevSecOps, CI/CD security gating, or SAST/DAST tooling
• Prior technical content review, assessment design, or QA for security-focused engineering tasks
Sourced from Mercor · original listing · application link last checked 31 Aug 2026
Design and craft multi-turn conversational datasets to improve AI agent reasoning, function calling accuracy, and real-world interaction capabilities.
Evaluate AI model outputs for accuracy, safety and helpfulness across domains.
Evaluate personalization quality in AI systems using cultural and contextual understanding of Japanese user behavior.
Tell us what you know — we'll surface the AI training work that fits.