AI Red Teaming Engineer
About the Role:
We are seeking an experienced AI Red Teaming Engineer to proactively identify vulnerabilities, safety gaps, and abuse scenarios within AI-powered systems. The role involves simulating attacker behavior to evaluate whether AI models can be manipulated into bypassing safety controls, violating policy restrictions, exposing sensitive information, or performing actions outside their intended design.
Key Responsibilities
Adversarial Testing:
Design and execute AI red team assessments.
Identify methods to bypass AI safety guardrails.
Test prompt injection, indirect prompt injection, and jailbreak techniques.
Evaluate resistance to social engineering and manipulation attacks.
Assess model adherence to authorization and permission boundaries.
AI Abuse Simulation:
Develop test cases for content restriction bypasses.
Test unauthorized data access and sensitive information disclosure.
Evaluate harmful content generation scenarios.
Test privilege escalation and prompt chaining attacks.
Perform multimodal attacks involving text, images, and documents.
Security Assessment:
Identify weaknesses across AI workflows.
Perform threat modeling exercises.
Test AI integrations with enterprise systems and tools.
Evaluate secure handling of confidential information.
Validate alignment with Responsible AI and security standards.
Risk Analysis:
- Document vulnerabilities, attack paths, and findings.
- Assign severity based on potential business impact.
- Recommend mitigation and security improvements.
- Partner with engineering teams to validate remediation.
- Collaboration Work closely with teams across:
- Responsible AISecurity Engineering
- Trust & Safety
- PrivacyProduct Security
- Infrastructure Security
- Applied Research
Required Qualifications:
Experience working with Large Language Models (LLMs).Knowledge of AI security and adversarial testing.
Strong understanding of prompt engineering.
Familiarity with threat modeling methodologies.
Strong cybersecurity fundamentals.
Experience identifying and documenting security vulnerabilities.
Preferred Background: Candidates with experience in one or more of the following areas are preferred:
- AI Red Teaming
- Security Research
- Red Team Operations
- Trust & SafetyAI Safety
- Adversarial Machine Learning
- Penetration TestingLLM Security
- Key Skills / KeywordsAI Red Teaming, Prompt Injection, Jailbreak Testing, Adversarial AI, Responsible AI, Security Research, Trust & Safety, AI Risk Assessment, LLM Security, AI Resilience, Threat Modeling
- Experience LevelWe are open to a mix of experience levels, including:
- Less than 5 years5-10 years 10+ years