AI Jailbreak & Prompt-Injection Security Expert

About OpenTrain

OpenTrain is the centralized platform where people build careers in AI training and data labeling. We connect trained contributors with specialized projects, help you consolidate your work across the industry, and let you build a unified AI training portfolio you control.

Creating an OpenTrain account is free and gives you direct access to contract opportunities where your testing and evaluation work shapes how AI behaves.

• OpenTrain AI is the hiring and contracting organization for this role.
• You will work remotely for a flexible, project-based engagement.

About AI training work

AI training (also called data labeling or human feedback work) is the human side of building modern AI systems: people prepare examples, test model behavior, and evaluate outputs so models learn to be safe and useful.

This class of work is often remote, flexible, and accessible — contributors directly influence model safety, robustness, and real-world behavior.

• 100% remote: work from anywhere with a computer and internet.
• Flexible, part-time work that can fit around other commitments.
• Your evaluations and red-team findings help improve state-of-the-art AI systems.

The role

OpenTrain AI is recruiting an AI jailbreak and prompt-injection security expert to evaluate model safety, red-team LLM behavior, and uncover adversarial bypass patterns. You will design evaluation methodologies, build regression tests, and translate security findings into clear recommendations.

This is a part-time contractor role intended for contributors who can commit 20+ hours per week; the position is remote and open worldwide. Language of work: English.

• Engagement type: Contractor, part-time.
• Time commitment: 20+ hours per week.
• Work language: English; candidates worldwide are eligible.

What you'll do

Your day-to-day work focuses on adversarial testing and documenting vulnerabilities so engineering and safety teams can reproduce and remediate issues.

• Design methods for evaluating AI system safety and robustness.
• Test models for ethical jailbreaks, prompt injection, and tool-use abuse.
• Build and maintain regression suites for adversarial vulnerability checks.
• Develop cross-domain elicitation strategies for multi-turn bypass patterns.
• Document findings, methodologies, and best practices for technical and non-technical audiences.

Requirements

Candidates must have practical experience in adversarial ML, LLM red teaming, or closely related AI security/testing work and be able to clearly document and communicate findings.

The role accepts a range of backgrounds but requires demonstrated hands-on testing experience with prompt injection, ethical jailbreaks, or tool-use abuse.

• Experience in adversarial machine learning, LLM red teaming, or AI safety evaluation.
• Hands-on testing experience with ethical jailbreaks, prompt injection, or tool-use abuse.
• Familiarity with current LLM architectures, prompt engineering techniques, and security assessment tools.
• Strong written and verbal communication skills (able to explain technical findings to non-technical audiences).
• Advanced degree in computer science, cybersecurity, machine learning, or equivalent professional background is preferred.

Nice-to-have

You’ll stand out if you’ve contributed to the AI security community through research, tooling, or public presentations.

• Published research, open-source tools, or conference presentations in AI security or adversarial ML.
• Experience on cross-functional AI safety or security projects.

Compensation, data type, and labeling details

This is an hourly paid, per-hour engagement. Work focuses on text-based red-teaming and evaluation labeling tasks.

• Payment type: PAY_PER_HOUR, USD $50–$90 per hour (hourlyRate listed as $90 with range $50–$90).
• Data type: TEXT; Labeling tasks: RED_TEAMING and EVALUATION_RATING.
• Employment types: CONTRACTOR, PART_TIME.

How to apply and next steps

If this role matches your skills and availability, prepare examples of past red-team work, adversarial tests, or relevant research to include with your application. Clear, reproducible documentation and example test cases will help reviewers evaluate fit.

OpenTrainAI will review applications and coordinate the contracting process. Creating or updating your OpenTrain profile is the first step to apply and be considered.

• Provide sample reports, test cases, or links to tools/research when you apply.
• Be prepared to demonstrate practical testing approaches and walk through a documented finding.