Remote Cybersecurity AI Trainer and Evaluation Specialist

agency; United States; Remote; USD 8,300 - 90,000; Full time

Job summary

agency is seeking a freelancer to leverage cybersecurity expertise to evaluate how advanced language models respond to security-focused prompts and scenarios. You will assess factual accuracy, technical soundness, and logical reasoning across topics from network security to incident response, documenting weaknesses for model improvements.

The role emphasizes clear step-by-step reasoning and the ability to report reproducible error traces, flexibly remotely, with compensation described as

Qualifications

• Strong cybersecurity knowledge and the ability to assess security guidance critically.
• Clear communication and the ability to explain reasoning step by step.
• A degree in cybersecurity, computer science, information assurance, or a related field is preferred.
• Relevant certifications or practical experience in security operations, red or blue team work, or cyber policy are useful indicators of fit.

Responsibilities

• Work through cyber incident scenarios and theoretical security questions with a language model.
• Review answers for factual accuracy, technical soundness, and logical consistency.
• Test topics such as network security, ethical hacking, digital forensics, threat intelligence, cryptography, malware analysis, vulnerability assessment, and incident response.
• Explore penetration testing, intrusion detection, security architecture, zero trust, ransomware mitigation, secure coding, and risk management.
• Record reproducible error traces and failure patterns; suggest improvements to prompts and evaluation criteria.

Education

Degree in cybersecurity

Job description

agency is seeking a freelancer to leverage cybersecurity expertise to evaluate how advanced language models respond to security-focused prompts and scenarios. You will assess factual accuracy, technical soundness, and logical reasoning across topics from network security to incident response, documenting weaknesses for model improvements.

The role emphasizes clear step-by-step reasoning and the ability to report reproducible error traces, flexibly remotely, with compensation described as