Break it
Prompt injection (direct and indirect), jailbreaks, data exfiltration through tools, model extraction and abuse of autonomous agents, mapped to MITRE ATLAS.
RECRUITMENT AI security
AI security recruitment for the people who secure the machines that think.
Every organisation is shipping AI. Very few people know how to break it, secure it and govern it. We find them: AI red teamers, LLM security engineers and AI governance leaders across Australia.
- OWASP LLM Top 10
- MITRE ATLAS
- ISO/IEC 42001
- NIST AI RMF
AI & ML security specialists01 Roles
The AI security roles that matter now.
02 The skills
What a real AI security specialist knows.
We assess candidates across the three sides of the discipline, so you hire for the job you actually have.
Secure it
Threat modelling RAG pipelines and agents, guardrails, least-privilege tool design, evaluation harnesses, secure model hosting and MLSecOps.
Govern it
AI risk registers, ISO/IEC 42001 management systems, the NIST AI RMF, Australian AI guidance and privacy obligations for automated decisions.
03 How we find them
We map skills, not job titles.
Most AI security experts don't have "AI security" in their title. They're pen testers who red-team chatbots, AppSec engineers who threat model agents, or ML engineers who built the guardrails. Our AI searches for the evidence of those skills, and our practitioners test them.
Shortlisted candidates take a Critical Hive interview built from your role, so you see how they reason through a real AI attack scenario before you meet them.
SAMPLE INTERVIEW QUESTION
"A product team plans to launch an AI agent that reads customer emails and can issue refunds. How would you threat model it before launch, and what are your launch conditions?"
Scored against: trust boundaries, indirect prompt injection, tool permissions, human approval, evaluation before go-live.
04 FAQ
AI security hiring, answered.
Read our guide: How to hire an AI security specialist
What is an AI security specialist?
An AI security specialist protects systems that use machine learning and large language models. The work covers attacking AI (red teaming, prompt injection, jailbreaks, data leakage and tool misuse), securing it (threat modelling RAG pipelines and AI agents, guardrails, MLSecOps) and governing it (ISO/IEC 42001, the NIST AI RMF and Australia's AI guidance).
Which AI security roles do you recruit for?
AI red teamers, LLM security engineers, AI security architects, MLSecOps engineers, adversarial ML researchers, model risk analysts, AI governance leads, AI assurance consultants and Heads of AI Security.
Why is AI security talent so hard to find?
The discipline is only a few years old. Most strong candidates come from offensive security, application security or ML engineering and have built AI security skills on the job, so they rarely have a matching job title. Keyword searches miss them. We find them by mapping skills, not titles.
How do you test an AI security candidate's skills?
With scenario-based interviews: for example, asking how they would red-team an assistant that reads company documents and calls internal APIs. Answers are scored against the OWASP Top 10 for LLM applications, MITRE ATLAS and real engagement experience, then reviewed by a practitioner.
Do you also offer AI security consulting?
Yes. If you need expertise before a hire is in seat, our AI security advisory covers AI red teaming, ISO 42001 governance and secure AI architecture reviews.
WE USE THIS TOO
We screen these roles with our own AI interview platform.
Every shortlist on this page passes through Critical Hive first: an AI interview, evidence-based scoring, integrity checks. Run it on your own roles — 8 interviews free.