AI Red Teamer: Adversarial Security for AI Systems Remote

10a Labs; Boston (MA); Remote; USD 100,000 - 120,000; Full time

Job summary

10a Labs is seeking an AI Red Teamer with deep cybersecurity expertise to evaluate the security and resilience of advanced AI systems. You will design red team assessments, craft adversarial inputs, and collaborate with engineers to improve system safety.

The role emphasizes offensive security, creative problem solving, and thinking like an adversary, with travel and in-person onboarding as described.

Qualifications

• 2–5 years of relevant professional experience in security.
• Strong understanding of cybersecurity principles and attack methodologies.
• Experience with security tools and Linux environments.
• Ability to translate objectives into structured testing plans.

Responsibilities

• Design and execute red team assessments against AI systems and infrastructure.
• Develop adversarial testing methodologies to evaluate security and resilience.
• Craft prompts and adversarial inputs to uncover unsafe system behavior.
• Identify vulnerabilities and communicate actionable risk and fixes.
• Collaborate with engineers and researchers to improve system security.
• Document findings and track remediation across teams.
• Refine testing methodologies as threats evolve.
• Translate complex results into clear executive summaries.

Skills

Cybersecurity

Python scripting

Linux OS

SIEM basics

Analytical thinking

Threat modeling

Education

Degree in Computer Science, Engineering, Cyber Security, or related field

Tools

Splunk

Wireshark

Metasploit

Burp Suite

Nmap

Job description

10a Labs is seeking an AI Red Teamer with deep cybersecurity expertise to evaluate the security and resilience of advanced AI systems. You will design red team assessments, craft adversarial inputs, and collaborate with engineers to improve system safety.

The role emphasizes offensive security, creative problem solving, and thinking like an adversary, with travel and in-person onboarding as described.