AI Red Team Engineer

Arcitix Security; United States; On-site; USD 100,000 - 140,000; Full time

Job summary

Arcitix Technologies is seeking an AI Red Team Engineer to proactively find vulnerabilities in AI systems. Your responsibilities will include designing tests, developing new techniques for security, and creating reports that help clients improve their systems.

The role requires strong offensive security experience, Python skills, and the ability to communicate technical findings clearly. Ideal candidates will have hands-on experience with LLMs and machine learning security. Join us in shaping the future of AI safety.

Qualifications

• 3+ years in offensive security, ML engineering, or applied AI research.
• Hands-on experience attacking or defending LLM applications.
• Strong Python skills and fluency with modern LLM tooling.

Responsibilities

• Design and execute adversarial tests against LLMs, RAG systems, and agents.
• Develop novel jailbreak, prompt-injection, and data-exfiltration techniques.
• Write clear technical reports with reproducible steps and remediation guidance.

Skills

Offensive security

ML engineering

Python

LLM tooling

Communication

Tools

LangChain

LlamaIndex

MCP

Job description

Break LLMs, agents, and ML systems before attackers do — jailbreaks, prompt injection, data exfiltration, and abuse chains.

About the role

As an AI Red Team Engineer at Arcitix, you spend your days finding the failure modes in production AI systems before adversaries do. You will design adversarial campaigns against LLM applications, autonomous agents, and ML pipelines, then translate what you break into concrete, prioritized fixes our clients can ship.

This is a hands-on, deeply technical role for someone who thinks like an attacker and writes like an engineer. You will work across client engagements and internal tooling, and your findings will directly shape our LLM firewall and detection products.

Key Responsibilities

• Design and execute adversarial tests against LLMs, RAG systems, and agents
• Develop novel jailbreak, prompt-injection, and data-exfiltration techniques
• Map every finding to the OWASP LLM Top 10 and MITRE ATLAS
• Build internal tooling to automate and scale AI red-team campaigns
• Write clear technical reports with reproducible steps and remediation guidance

Requirements

• 3+ years in offensive security, ML engineering, or applied AI research
• Hands-on experience attacking or defending LLM applications
• Strong Python skills and fluency with modern LLM tooling
• Working knowledge of the OWASP LLM Top 10 and MITRE ATLAS
• Ability to communicate complex findings to both engineers and executives

Nice to have

• Published research or CTF experience in AI/ML security
• Experience with agent frameworks (LangChain, LlamaIndex, MCP)
• Background in traditional application security or red teaming