AI Safety & Red Teaming Specialist
$104,000–$187,200 year
RemoteUnited States or India
Job Summary
Design and implement advanced evaluation methodologies for AI system safety, including ethical jailbreak testing, prompt injection detection, and LLM red teaming. Develop cross-domain adversarial testing strategies to uncover complex, multi-turn attack patterns and model vulnerabilities. Build, maintain, and enhance regression test suites to continuously assess jailbreak susceptibility and prompt injection risks. Create comprehensive evaluation frameworks that simulate real-world adversarial threats to improve AI robustness and reliability. Collaborate with technical teams to translate security findings into actionable recommendations for AI safety improvements. Document testing methodologies, findings, and best practices through clear technical reports and presentations for both technical and non-technical stakeholders. $50-$90/hour, remote contractor role.
Required Qualifications
- 2+ years of experience in AI Safety, Adversarial Machine Learning, LLM Red Teaming, AI Security, or a related field
- Hands-on experience researching, testing, or identifying vulnerabilities involving prompt injection, ethical jailbreaks, adversarial attacks, or tool-use exploitation
- Strong understanding of modern LLM architectures, prompt engineering, and AI safety evaluation methodologies
- Experience developing structured security assessments, regression testing frameworks, and adversarial evaluation strategies
- Excellent analytical, documentation, and communication skills with the ability to explain complex technical findings clearly
- Ability to collaborate effectively within cross-functional technical teams
Desired Qualifications
- Master's or PhD in Computer Science, Cybersecurity, Machine Learning, Artificial Intelligence, or a related discipline
- Contributions to AI security research, open-source AI safety tools, conference presentations, or published research
- Experience with AI model evaluation frameworks, prompt engineering techniques, and AI security assessment tools
- Background in multidisciplinary AI safety, cybersecurity, or adversarial machine learning projects
Hiring someone like this?
Get your role in front of qualified candidates on Sorce.