Adversarial Prompt Engineer

Adversarial Prompt Experts design and run targeted tests that probe large language models for failure modes and potentially harmful outputs. You will craft prompts and scenarios to evaluate model guardrails, attempt creative bypasses, and document results so engineers and safety researchers can improve defenses. Key Responsibilities

• Develop domain-specific prompts and adversarial scenarios to probe model behavior.
• Experiment across multiple LLMs, both open- and closed-source, to compare responses and identify weaknesses.
• Explore evasion techniques and jailbreaking approaches while maintaining ethical boundaries.
• Systematically log attempts, inputs, and outcomes, and produce clear reports of findings.
• Collaborate with engineers and safety researchers to communicate issues and suggest mitigations.
• Spend time researching topics relevant to your tests, using AI tools as appropriate to accelerate investigation., * Complete identity verification as part of eligibility checks.
• Browse available project listings, enroll in projects that match your skills, and complete onboarding steps for each project.
• Begin assigned work and receive payment according to the project terms.

• Hands-on experience using multiple large language models, with comfort experimenting across systems.
• Proven prompt engineering skills, including crafting prompts and applying evasion or jailbreaking techniques.
• An adversarial or security mindset, with experience in red teaming or offensive security considered a strong plus.
• Persistence and creativity, willing to iterate on many variations and push edge cases.
• Strong documentation skills, able to log attempts and summarize outcomes clearly for technical and non-technical audiences.
• Ethical awareness and responsibility when handling sensitive or harmful content., Job Requirements 10+ years of software engineering leadership experience focusing on AI implementation and large-scale process automation. Proficiency in Appian for architectin…

• Part-time, remote, largely asynchronous work from any location.
• Flexible hours, with an ideal commitment of 10+ hours per week.
• Some synchronous meetings will be scheduled and are highly recommended to support success on projects.
• Placement into specific projects depends on project availability and alignment with your skills.

Compensation

• Hourly rates start at $45 and extend up to $65 depending on prior experience with adversarial testing.
• Up to $65/hr (depending on the project)

Eligibility

• This opportunity is open to U.S.-based candidates and to recent graduates who have U.S. work authorization.
• F-1 students who are eligible for CPT or OPT may be eligible to participate. Consult your Designated School Official to confirm whether participation meets your school requirements. If your school requires enrollment in a CPT course, these projects may not qualify. STEM OPT is not supported. Refer to the program help resources for more details.