Penetration Tester (AI Training)

Overview

Models can describe an exploit chain in confident detail and miss that step three never actually works, the payload gets filtered, or the whole path assumes a misconfiguration no real target would have. Someone has to know which attacks land and which just sound good, and that someone is you.

As a Penetration Tester you'll evaluate how models reason about offensive security, break the chains that don't hold, and write the attack paths a real operator would take.

What you’ll actually do

• Test offensive reasoning across recon, exploitation, privilege escalation, and lateral movement, and see where the model bluffs.
• Break the exploit chains that don't hold up: filtered payloads, impossible preconditions, and steps that skip the hard part.
• Write the attack path that actually works when the model falls short, with the tradecraft a real operator would use.

Roles this fits

Common backgrounds: Penetration Tester, Red Team Operator, Offensive Security Engineer.

What we look for

• Hands-on offensive security experience: penetration testing, red-teaming, or CTF at a serious level.
• Fluency with the attacker's toolkit and how real exploit chains come together.
• Clear written English: your explanations are the training signal.
• No degree required. We care about what you can do, not where you learned it.

Compensation

Up to $125+/hr depending on task difficulty and specialization. Many contributors add $100k–$200k a year; some make it their full-time income.

About DataAnnotation

DataAnnotation is where 1M+ experts train the world’s leading AI models. Hundreds of millions paid to contributors to date, and the average contributor stays 5+ years. Flexible, remote, and always project-available.

Apply now

Hourly rate up to $125+ / hour

Commitment Flexible hours

Type Independent contractor

Payouts Weekly

Apply now

<h2

Where you can work

This role is 100% remote, for applicants located in these 6 countries: Australia, Canada, United Kingdom, Ireland, New Zealand and United States.