Find Jobs in AI All opportunities
Open for applicationsCoding & data

AI Jailbreak & Prompt-Injection Security Expert — AI Evaluation

Open role
Pay
$50–$90 / hour
Location
Remote
Experience level
Expert

A CLOSER LOOK

Explore the opportunity

01 / THE DETAILS

Role overview

Get to know the opportunity

Apply practical AI security evaluation expertise to specialist AI training and evaluation work.

02 / THE DETAILS

What you’ll be doing

Your day-to-day contribution
  • Assess authorised jailbreak, prompt-injection, tool-use-abuse and adversarial-testing scenarios.
  • Identify technical errors or missing evidence and explain their significance.
  • Write clear, structured findings and review project feedback.
03 / THE DETAILS

Who this could suit

See where your experience could fit

Experienced professionals able to assess realistic work and explain their reasoning clearly.

04 / THE DETAILS

What you’ll need

Experience, skills and essential requirements
  • At least two years in adversarial ML, LLM red teaming, AI safety or a related security area is stated.
  • An advanced relevant degree or equivalent professional/operational experience is accepted.
  • You need a demonstrated vulnerability-research record and strong documentation.

You don’t have to match every requirement exactly. We welcome applicants with different backgrounds, levels of experience and transferable skills.

05 / THE DETAILS

Pay and working arrangements

Rates, commitment and working arrangements

Remote contractor work paying $50–$90 per hour. Weekly hours, duration and payment dates are not stated.

06 / THE DETAILS

What happens next

How to move forward

Apply through Find Jobs in AI with your profile and CV. We’ll review your application and, if you’re a good fit, we’ll be in touch about the next steps.

Practice for this role

Review a specialist conclusion

A quality report required checks on five samples. Four passed, and the fifth was not measured. The draft says “All five samples passed, so the batch meets the release rule.”

Explain what the evidence supports and what you would ask the team to check before release.

Sign in for private feedback or work through the example below.

See the rubric and example answer
  • Evidence · Four passed and the fifth has no recorded result.
  • Conclusion · The evidence does not establish that all five passed or that the release rule was met.
  • Follow-up · Measure and document the fifth sample before deciding.

Four samples passed. There is no result for the fifth, so the release rule has not been verified. Measure and document the fifth sample before deciding.

Application checklist
Requirements from the listing

At least two years in adversarial ML, LLM red teaming, AI safety or a related security area is stated. An advanced relevant degree or equivalent professional/operational experience is accepted. You need a demonstrated vulnerability-research record and strong documentation.

Sign in to check your saved CV.

Pay$50–$90 / hour
Continue to apply