What an AI Security & Red Teaming Specialist does
The AI Security & Red Teaming Specialist role consists of attacking your AI systems the way a real adversary would (prompt injection, jailbreaks, data exfiltration through completions, agents tricked into misusing their tools), then turning every finding into guardrails and regression evals. Classic pentesting checks your infrastructure; this role checks your model's behavior. If your AI touches customers, money, or sensitive data, someone will eventually run these attacks. Better if you paid them to.
Monthly rate
$7,000–$10,000/mo
All-in: contract, benefits, equipment, IP
Experience
10+ years security
2+ in AI red teaming
Location
Latin America
Argentina · Colombia · Mexico
Timezone
Full US overlap
Fluent English, onboarded in one week
Core stack
AI tools, daily
Verticals seen
What an AI Security & Red Teaming Specialist owns - and what they don't
What they own
- Run adversarial audits on LLM features and agents: injection, jailbreaks, exfiltration paths, tool misuse
- Test agentic systems for privilege escalation: can an agent be talked into touching what it shouldn't?
- Review the AI attack surface: data flows, credential models, tenant isolation, what leaves the perimeter
- Turn findings into working defenses: guardrails, filters, and adversarial eval suites wired into CI
- Re-test after every significant model, prompt, or tool change: the surface reopens each release
What they don't - and who does instead
- Design the agent architecture. That's the AI Agent Architect; this role tries to break it
- Build the product features being tested. That's AI Engineering
- Replace your infrastructure pentest or SOC 2 audit. This is a complementary surface, not a substitute
- Rubber-stamp launches. If it's not safe to ship, the report says so
Who hires an AI Security & Red Teaming Specialist, and for what
Fintech and healthcare shipping customer-facing AI. Regulated data plus generative output is the highest-stakes combination there is. They audit before launch because the alternative is explaining a leak to a regulator.
Companies deploying agents with real permissions. Once an agent can write to systems or move money, 'what could someone make it do?' becomes a board-level question that needs a tested answer.
AI product companies selling to enterprises. Enterprise security teams now ask for AI-specific testing evidence in procurement. A red-team report is becoming table stakes to close.
- 01
Pre-launch adversarial audit. Full attack pass on a customer-facing assistant or agent before it meets real users: findings, reproducible cases, fixes.
- 02
Agentic platform security review. Credential isolation, tool permission boundaries, cross-tenant leakage: the specific ways multi-agent systems fail.
- 03
Standing adversarial evals. Attack cases wired into CI so every prompt or model change gets red-teamed automatically, not just at launch.
- 04
Procurement and compliance evidence. The AI-specific security documentation enterprise buyers and auditors now expect to see.
Work our engineers at this role have shipped
- Pre-launch adversarial audit of a customer-facing LLM assistant: injection, exfiltration, and off-policy behavior mapped and closed
- Security review of a multi-tenant agentic platform: credential isolation, tool permission boundaries, cross-tenant leakage tests
- Standing adversarial eval suite added to a client's CI so every model or prompt change gets red-teamed automatically
Do you actually need an AI Security & Red Teaming Specialist?
You do, if:
- Your AI feature is customer-facing and handles data you'd have to disclose losing
- Agents in your stack hold credentials or tool access an attacker would want
- An enterprise prospect's security questionnaire has an AI section you can't answer yet
- Nobody has ever seriously tried to prompt-inject your own system
You probably don't, if:
- Your AI is internal-only, read-only, on non-sensitive data. Basic guardrails from your AI Engineer may be enough for now
- You haven't built anything yet. Bring this role in before launch, not before the prototype
- What you need is general infra security. Hire a classic security engineer first
Not sure which role fits? Tell us the problem instead of the title - we'll tell you what we'd actually staff, even if it's not this. If it is this: discovery call today, matched profiles in 48 hours, onboarded in a week.
Need more than one role, or a whole team around this hire? See how nearshore staff augmentation works. Building this across your whole product, not just one hire? See our AI engineering service.
Hire a Senior AI Security & Red Teaming Specialist