Cybersecurity Specialist for AI Model Training

Cybersecurity experts apply their technical and offensive security experience to evaluate and improve large language model outputs on security-related tasks. You will contribute subject-matter knowledge to projects that create security-focused prompts and assess LLM responses for accuracy, safety, and relevance. The program runs year-round, although specific project availability varies by domain.

• Develop domain-specific prompts, scenarios, and test cases for LLM evaluation
• Assess model responses for technical correctness, exploitable guidance, and adherence to safety constraints
• Provide detailed, structured feedback and examples to guide model improvement
• Document results and contribute to evaluation datasets and reporting

• At least 3 years of professional experience in security engineering or offensive security
• Relevant roles include Security Engineer, Penetration Tester, Red Team Engineer, Malware Analyst, Threat Hunter, Threat Intelligence Analyst, Offensive Security Engineer, Vulnerability Researcher, or Security Consultant
• Hands-on experience with one or more of the following tools: Burp Suite, Nmap, Wireshark, Metasploit Framework, BloodHound, Impacket, Nessus, or Hashcat
• Openness to learn new skills while contributing to AI evaluation across multiple disciplines

• Location: Remote, fully asynchronous work
• Employment type: Part-time
• Flexible hours, with no minimum weekly commitment required
• Placement on specific projects is dependent on project availability; projects appear throughout the year by domain

• Up to $65/hr (depending on the project)

• F-1 students with CPT or OPT may be eligible to participate, subject to your school and Designated School Official guidance
• If your institution requires enrollment in a CPT course or other course-based requirement, this program may not satisfy that requirement
• STEM OPT is not supported for this work
• Refer to the program help resources for detailed information on supported work authorizations

• Create an account with the program and complete your profile
• Complete identity verification as required
• Join or be matched to relevant projects and complete any project-specific onboarding
• Begin contributing to evaluations and receive payment for completed project work