Remote AI Red Team Analyst - Adversarial Security Expert

Alignerr Corp.; United States; Remote; USD 83,000 - 165,000; Part time

Job summary

Alignerr seeks an AI Red Team Analyst to stress-test and harden AI systems by probing guardrails, crafting adversarial prompts, and delivering structured findings that shape safety for millions of users.

This rare role sits at the intersection of cybersecurity and cutting-edge AI. It offers remote hourly work with autonomy to set your own hours and weekly commitment, collaborating with researchers and engineers worldwide.

Qualifications

• Solid grasp of cybersecurity concepts including threat modelling, pentesting, or vulnerability research.
• Experience with AI/ML systems, LLMs, or prompt engineering is beneficial.
• Ability to document findings clearly so others can act on them.

Responsibilities

• Design and conduct red-teaming exercises to uncover security weaknesses in AI systems
• Craft adversarial prompts, jailbreak attempts, and edge-case scenarios to challenge model guardrails
• Evaluate AI outputs for safety violations, bias, and policy compliance
• Document vulnerabilities, exploits, and unexpected behaviours in clear, structured reports
• Collaborate with engineering teams to recommend practical mitigations and improvements
• Stay sharp on emerging AI security threats, attack techniques, and evolving best practices
• Help define and refine security evaluation rubrics and testing protocols

Skills

Threat modeling

Penetration testing

Vulnerability research

AI/ML systems

Prompt engineering

Clear reporting

Asynchronous work

Job description

Alignerr seeks an AI Red Team Analyst to stress-test and harden AI systems by probing guardrails, crafting adversarial prompts, and delivering structured findings that shape safety for millions of users.

This rare role sits at the intersection of cybersecurity and cutting-edge AI. It offers remote hourly work with autonomy to set your own hours and weekly commitment, collaborating with researchers and engineers worldwide.