AI Red Team Security Engineer

Ethos Life

2h•$152,000 - $269,000•Remote

About The Position

Ethos is a leading life insurance technology company on a mission to protect families by democratizing access to life insurance and empowering agents at scale. With its robust three-sided technology platform, Ethos is transforming the life insurance experience for consumers, agents, and carriers alike. Ethos offers instant, accessible products and a seamless online process that requires no medical exams and just a few health questions; it eliminates traditional barriers, making it easier than ever for everyone to protect their families. Ethos is redefining how life insurance is bought, sold, and underwritten. We are looking for a skilled and creative AI Red Team Engineer to join our offensive security team. In this role, you will simulate real-world adversaries, exploit vulnerabilities across applications, cloud infrastructure, and AI/ML systems using both traditional penetration testing techniques and cutting-edge AI-augmented attack tooling. You will operate across the full attack surface: web apps, APIs, mobile, internal networks, and AI-powered products including LLM pipelines, model APIs, agents, and RAG systems. You will help us find the flaws before the adversaries do, and work closely with engineering and product teams to close those gaps.

Requirements

7+ years of hands-on penetration testing and offensive security experience in a professional setting
Demonstrated experience testing AI/ML systems, LLM-powered products, or AI APIs
Experience conducting red team engagements
Scripting and tool development
Strong understanding of authentication protocols and common implementation flaws
Familiarity with cloud security architectures and common misconfigurations
Working knowledge of Docker/Kubernetes and container security
Understanding of LLM architectures and how they relate to attack surfaces.
Familiarity with OWASP LLM Top 10
Practical experience with prompt injection and jailbreak techniques against LLMs
Ability to use LLMs as force-multipliers in red team workflows

Nice To Haves

Certifications: OSCP, OSEP, CRTO, CRTE, PNPT, CEH, GPEN, GWAPT, or equivalent
Experience with adversarial ML frameworks
Contributions to open-source security tooling or published CVEs / bug bounty hall-of-fame credits
Familiarity with AI governance frameworks
Experience with GenAI infrastructure
Background in threat modeling for AI-powered applications
Reverse engineering skills for binary and mobile assessments
CTF participation or competitive hacking experience

Responsibilities

Design and execute adversarial attacks against large language model (LLM)-powered products including prompt injection, jailbreaking, goal hijacking, and context manipulation.
Test retrieval-augmented generation (RAG) pipelines for data exfiltration, poisoning, and unauthorized knowledge extraction.
Assess AI agent systems and agentic workflows for unsafe tool-use, privilege escalation, and indirect prompt injection via environment feedback.
Conduct model extraction, membership inference, and adversarial example attacks against deployed ML models.
Evaluate AI guardrails, safety filters, and content moderation layers for bypass techniques.
Perform full-scope penetration tests across web applications, REST/GraphQL APIs, mobile apps (iOS/Android), cloud environments (AWS, GCP, Azure), and internal networks.
Conduct red team exercises simulating advanced persistent threat (APT) actors using MITRE ATT&CK and AI-augmented techniques.
Exploit vulnerabilities across the OWASP Top 10 and beyond: SSRF, IDOR, XXE, SSTI, authentication bypasses, and logic flaws.
Perform social engineering and phishing simulations as part of combined red team campaigns.
Conduct cloud and Kubernetes security assessments including IAM misconfigurations, container escapes, and privilege escalation paths.
Leverage AI models and tools (e.g., LLMs, code generation, fuzzing assistants) to accelerate vulnerability discovery, payload crafting, and exploit development.
Build or adapt AI-powered reconnaissance, exploitation, and evasion tooling for internal use in red team engagements.
Stay current with adversarial AI research and translate academic findings into practical red team techniques.
Use AI to automate repetitive testing tasks and generate novel attack variants at scale.