Skip to content
SaidGig
Sign up.

Give me your email, I promise I won't do anything weird with it.

Adversarial Prompt Engineer

Up to $65/hr (depending on the project)

Remote — US onlyPart-timetechnology
Apply Now

About this role

Role Overview

Adversarial Prompt Experts design and execute prompts and scenarios that probe large language models for failure modes and harmful outputs. You will think like an adversary to discover guardrail weaknesses, develop creative evasion techniques, and work with engineers and safety researchers to surface findings and improve model defenses.

Key Responsibilities
  • Craft domain-specific prompts and scenarios to test LLM behavior and safety limits.
  • Probe model guardrails, explore bypass techniques, and iterate on prompt variants to uncover edge cases.
  • Systematically document attempts, inputs, model outputs, and observed failure modes.
  • Collaborate with engineers and safety researchers to communicate findings and suggest mitigations.
  • Spend time researching topics relevant to assigned projects, using AI tools as an aid when appropriate.
  • Participate in occasional synchronous meetings to align with project teams, while completing most work asynchronously.
Qualifications
  • Substantial hands-on experience using multiple LLMs, including both open-source and closed models.
  • Proven skill in prompt engineering, including experience with jailbreaking and evasion techniques.
  • An adversarial or security mindset, with bonus points for red teaming or offensive security experience.
  • Persistence and creativity, comfortable testing many variations and pushing edge cases.
  • Strong documentation habits, able to log experiments and communicate issues clearly.
  • High ethical awareness, able to handle sensitive content responsibly.
Work Terms
  • Part-time, remote role with largely asynchronous work.
  • Flexible hours, ideally able to commit 10+ hours per week.
  • Some synchronous meetings are highly recommended to ensure project success.
  • Work is project-based, and placement onto specific projects depends on project availability.
Compensation
  • Hourly rates start at $45 and extend up to $65 depending on prior experience with adversarial testing.
  • Displayed maximum: Up to $65/hr (depending on the project)
Eligibility
  • Open to U.S.-based candidates and recent graduates with U.S. work authorization.
  • F-1 students who are eligible for CPT or OPT may be eligible, subject to confirmation by your Designated School Official.
  • If your school requires enrollment in a specific CPT course, this engagement may not meet that requirement.
  • STEM OPT is not supported.
Application Process
  • Create an account and complete your profile.
  • Complete any required identity verification steps.
  • Enroll in onboarding for projects that match your skills and interests.
  • Complete onboarding, begin project work, and receive payment for completed assignments.

Related Jobs