AI Jailbreak & Prompt-Injection Security Expert

$50–90/hr

rate as listed

Workload: ~15h/week
Engagement: Project-based
Location: Worldwide
Field: STEM and Coding

Design adversarial evaluations for jailbreaks, prompt injection, tool-use abuse and other LLM security failures. The work includes creating reproducible regression tests, documenting findings and helping technical stakeholders turn them into model-safety improvements.

The work

Day to day, this contractor will design authorized adversarial tests for AI systems, including ethical jailbreaks, direct and indirect prompt injection, multi-turn bypasses and tool-use abuse. The work includes building cross-domain elicitation strategies and testing whether models follow malicious instructions, misuse connected tools or expose information under realistic attack conditions.

You will turn successful and unsuccessful attacks into maintainable regression suites, define evaluation frameworks, assess recurring failure patterns and update tests as systems change. Findings must be documented clearly enough for technical teams to reproduce them and for non-technical stakeholders to understand the risk and proposed mitigation.

What micro1 screens for

The screen is likely to focus on verifiable examples of adversarial ML, LLM red teaming or related security work, followed by detailed questions about methodology, evidence and evaluation judgment. The listing says prior AI experience is not required, but it prefers 2+ years in a relevant security domain, familiarity with current LLM architectures and prompt engineering, and evidence such as research, open-source tools or conference presentations. An advanced degree in computer science, cybersecurity, machine learning or a relevant discipline is preferred, with equivalent operational or professional experience also accepted.

Logistics

This is a remote contractor role. Async arrangements, typical weekly hours, project duration and required time-zone overlap are not specified, so candidates should be ready to state their realistic availability during screening. The platform presents the following as an observed band, not a guarantee of compensation or assignment volume.

Pay band: $50–90/hr

What they look for

Ethical Jailbreaks
LLM Red Teaming
Prompt Injection
Tool-Use Abuse

Included with this listing

Interview prep — 10 questions

Matched to what this platform's screening actually measures. Answer each out loud in 60–90 seconds: claim → concrete example → implication.

What it tests: This tests whether you can provide a traceable security example with a defined target, attack path, observable failure and reproducibility evidence.
A strong answer: Use one concrete case and state the system boundary, threat model, exact attack sequence, expected safe behavior, observed behavior and retest method. Clarify your personal contribution and distinguish confirmed evidence from hypotheses.
Where people lose it: Candidates lose points by naming attack categories or sharing a dramatic prompt without showing that the behavior was repeatable, in scope and security-relevant.

Summary, eligibility guidance and screening preparation are prepared by Sidequest from the platform's posting. Check the source listing before applying.

STEM AND CODING on Sidequest

What stem and coding AI work pays

Rates by specialty, the platforms hiring, and every open role in this field.