Quick overview
Job description
Job description
Cybersecurity SME, AI Red Team to apply offensive security expertise to evaluating AI models and coding agents for cybersecurity risks and unsafe capabilities.
In this role, you'll test AI models and coding agents in sandboxed environments to identify vulnerabilities and assess their ability to generate harmful offensive cyber tools. You'll evaluate findings against established safety specifications, develop testing methodologies, and provide actionable recommendations to engineering and safety teams. The role will also expand into agent and harness red-teaming as evaluation capabilities evolve.
The ideal candidate has strong offensive security experience, hands-on knowledge of exploit development and malware, and an understanding of AI models and agents. They're analytical, creative, and skilled at evaluating technical risks, documenting findings, and communicating complex security issues across teams.
As a Cybersecurity SME, AI Red Team , you'll:
- Red-team AI coding agents in sandboxed environments, using iterative testing techniques to evaluate whether models can generate offensive tools, including exploits, malware, and attack tooling.
- Assess model outputs against safety specifications, determine violations and severity, and identify when further testing is needed.
- Define and refine criteria for unsafe capability uplift, identifying ambiguous cases and gaps in existing safety specifications.
- Evaluate model capabilities in vulnerability discovery, exploit development, and malware analysis, verifying whether generated tools function as intended.
- Investigate agent-specific vulnerabilities, including prompt injection, tool misuse, and privilege escalation, as testing expands to agent and harness environments.
- Develop realistic attack scenarios and evaluation datasets, reviewing expert-written prompts and model outputs for technical accuracy.
- Document security findings, including severity, reproduction steps, and recommended mitigations, and communicate results to engineering and safety teams.
- Map findings to established frameworks, including MITRE ATLAS, OWASP Top 10 for LLM Applications, and NIST AI Risk Management Framework (AI RMF).
- Train and mentor red teamers and annotators on cybersecurity concepts while staying current on emerging AI attack techniques.
Qualification:
● Recognized certifications or proof of skill (OSCP, OSEP, GPEN, or published CVEs, CTF results, bug-bounty record).
● Prior AI/ML security, adversarial ML or model-evaluation experience.
● Experience in government, defense, intelligence or critical-infrastructure security.
● Malware reverse engineering, cloud security (AWS, Google Cloud Platform, Azure) or secure software development background.
● Conference talks, research papers or open-source security tooling.
Responsibilities
● Red-team coding agents in sandboxed environments: converse with the agent and use creative, iterative pressure to push the model into producing offensive tools such as exploits, malware and attack tooling.
● Judge each attempt against the spec: decide whether the output is a violation, whether it meets the bar, and whether to keep pushing. This judgment call is central to the role.
● Define and refine what counts as unsafe uplift, and flag gray areas and gaps in the spec for the team to resolve.
● Evaluate model capability on offensive cyber tasks such as vulnerability discovery, exploit development and malware analysis, and verify that generated tools actually work.
● Probe agent-specific weaknesses, including prompt injection, tool abuse and privilege escalation through agent actions, as our work moves toward harness and agent red teaming.
● Build realistic attack scenarios and evaluation datasets, and review expert-written prompts and model outputs for technical accuracy.
● Write clear findings with severity, reproduction steps and recommended mitigations, and brief engineering and safety teams on them.
● Map findings to frameworks such as MITRE ATLAS, OWASP Top 10 for LLM Applications and NIST AI RMF.
● Train and mentor red teamers and annotators on cybersecurity concepts, and keep the team current on new AI attack techniques.
Required skills
● 2+ years in offensive security: penetration testing, red teaming, vulnerability research or exploit development.
● Hands-on offensive tradecraft (exploit development, malware, post-exploitation, web and cloud attacks) and scripting ability (Python preferred), enough to judge whether model generated tools really work.
● Working understanding of how LLMs and agents work, and of common attack classes against them.
● Strong written communication: you can explain a technical risk to researchers and executives.
Pay: $65.00 - $70.00 per hour
Expected hours: 40.0 per week
Work Location: Remote
Will your resume get past this employer's screening? Check your resume against this job, free
Similar jobs
- 4C
Information Security AnalystRecently posted
4 Consulting Inc · United States · Hybrid · Posted 19 Nov 2026 🇺🇸
IAMTechnologySalary not listed - II
Vulnerability EngineerRecently posted
Innovative IT Solutions Inc · Malvern, PA · On-site · Posted 19 Nov 2026 🇺🇸
PythonJavaREST+1Technology$50/hr - CS

Product Security EngineerRecently posted
Cynet Systems · San Jose, CA · Hybrid · Posted 19 Nov 2026 🇺🇸
OWASPSOC 2Python+2Technology$85 - $90/hr - AG

AI Security EngineerRecently posted
AIT Global, Inc. · Princeton, NJ · Hybrid · Posted 19 Nov 2026 🇺🇸
GitLLMPython+2TechnologySalary not listed - LT

Security Administration AnalystRecently posted
Logisoft Technologies Inc · Jersey City, NJ · On-site · Posted 19 Nov 2026 🇺🇸
Microsoft ExcelData AnalysisAdministrative$50 - $60/hr - TS
Cyber Security ArchitectRecently posted
TechSpace Solutions Inc. · Juno Beach, FL · Hybrid · Posted 19 Nov 2026 🇺🇸
PCI DSSGDPRHIPAATechnologySalary not listed - DI

Senior CyberArk ArchitectRecently posted
Della Infotech · New York, NY · On-site · Posted 19 Nov 2026 🇺🇸
OnboardingHTMLLoad Balancing+6TechnologySalary not listed - MU

Cyber Security EngineerRecently posted
MphasiS Corporation USA · Atlanta, GA · Hybrid · Posted 19 Nov 2026 🇺🇸
EncryptionPCI DSSProject ManagementTechnologySalary not listed - DI

Senior CyberArk ArchitectRecently posted
Dizer Corp · New York, NY · Hybrid · Posted 19 Nov 2026 🇺🇸
Stakeholder ManagementOnboardingLoad Balancing+2TechnologySalary not listed - ID
Senior Application Security & Vulnerability Validation Engineer - Only W2Recently posted
Info Dinamica Inc · San Francisco, CA · On-site · Posted 19 Nov 2026 🇺🇸
AdministrativeSalary not listed - VS

AI Security ArchitectRecently posted
VKore Solutions LLC · United States · Hybrid · Posted 19 Nov 2026 🇺🇸
PythonJavaAWS+3TechnologySalary not listed - CI

Cyber Software Engineer
CACI International, Inc. · Florham Park, NJ · Hybrid · Posted 6 Nov 2026 🇺🇸
SciPyTCP/IPComputer Vision+5Technology$94.4k - $198.2k/yr
