LLM Red TeamJobs
LLM red team jobs hiring now. Find AI red teaming, model evaluation, prompt injection security, model abuse, agent sandboxing, and adversarial ML roles at AI and security companies.
LLM red teamers test language models, AI applications, and agent systems before attackers or abusive users find the weak points. The work spans prompt injection, jailbreak testing, model abuse scenarios, data exfiltration paths, tool-use failures, and sandbox escapes in agentic workflows.
Employers usually look for a mix of application security, Python, threat modeling, eval design, adversarial testing, and clear security writing. Strong candidates can turn messy findings into reproducible test cases, risk reports, mitigations, and regression evals that engineering teams can keep running.
Hiring comes from frontier AI labs, AI infrastructure startups, security vendors, crypto teams building agent systems, and companies deploying LLMs into sensitive workflows. Remote roles exist, especially for evaluation and AppSec-heavy work, but model labs and high-trust security teams often prefer hybrid or onsite collaboration.
Portfolio signals matter in this niche. Useful examples include public writeups on prompt injection, jailbreak taxonomies, eval harnesses, CTF or bug bounty results, adversarial ML projects, sandboxing research, and concise reports that show both exploit thinking and practical remediation.
Related Job Categories
Frequently Asked Questions
What does an LLM red teamer do?
LLM red teamers probe AI systems for unsafe behavior, security failures, and abuse paths. They test prompt injection, jailbreaks, tool-call misuse, data leakage, model abuse, and agent sandbox failures, then document reproducible findings and help teams build mitigations and regression evals.
What skills are needed for AI red teaming jobs?
Common requirements include application security, Python, threat modeling, prompt injection testing, eval design, adversarial testing, and strong written communication. ML knowledge helps, but many roles value security judgment and clear reporting as much as model internals.
Are LLM red team jobs remote?
Some LLM red team and model evaluation roles are remote-friendly because testing, reporting, and eval development can happen asynchronously. Roles at frontier labs, regulated companies, and teams handling sensitive unreleased models are more likely to require hybrid or onsite work.
How do I build a portfolio for LLM red teaming?
Publish concise technical writeups, build small eval harnesses, document prompt injection or jailbreak case studies, contribute to security tools, and show examples of findings translated into mitigations. Employers want evidence that you can test creatively and write clearly enough for engineering teams to act.