Remote AI Security Analyst — Red Team & Adversarial Testing

Alignerr Corp.; Greater London; Remote; GBP 83,000 - 165,000; Full time

Job summary

Alignerr Corp. is seeking security-minded professionals to probe, stress-test, and red-team AI models, uncovering weaknesses before they become real-world problems. You’ll design adversarial prompts, run tests on AI guardrails, and report findings to drive safer products used by millions.

This remote, flexible contract role lets you set your own schedule and work asynchronously, with 10–40 hours per week, focusing on threat modeling, penetration testing, and vulnerability research in AI systems.

Qualifications

• Experience in threat modeling, penetration testing or vulnerability research.
• Hands-on experience with AI/ML systems, LLMs, or prompt engineering.
• Self-directed and able to work asynchronously on task-based assignments.
• Strong written communication for clear reporting.

Responsibilities

• Conduct red-teaming exercises to surface security weaknesses in AI systems.
• Design and execute adversarial prompts and edge-case scenarios to test model guardrails.
• Evaluate AI outputs for safety, bias, and policy compliance.
• Document vulnerabilities, exploits, and unexpected behaviors in clear, structured reports.
• Collaborate with engineering teams to recommend practical mitigations and improvements.
• Stay sharp on emerging AI security threats, jailbreak 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

Technical reporting

Job description

Alignerr Corp. is seeking security-minded professionals to probe, stress-test, and red-team AI models, uncovering weaknesses before they become real-world problems. You’ll design adversarial prompts, run tests on AI guardrails, and report findings to drive safer products used by millions.

This remote, flexible contract role lets you set your own schedule and work asynchronously, with 10–40 hours per week, focusing on threat modeling, penetration testing, and vulnerability research in AI systems.