AI Red-Teamer

@ Crossing Hurdles

Overview

The AI Red-Teamer role at Crossing Hurdles involves red-teaming AI models, crafting exploits, and testing vulnerabilities. The position is remote with flexible scheduling and can be full-time or part-time.

Key Responsibilities

• Red-team AI models and agents by designing jailbreaks, prompt injections, misuse cases, and exploit scenarios.
• Generate high-quality human data by annotating AI failures, classifying vulnerabilities, and flagging systemic risks.
• Apply structured approaches using taxonomies, benchmarks, and playbooks to ensure consistent testing.
• Document findings comprehensively to produce reproducible reports, datasets, and attack cases.
• Support multiple projects including LLM jailbreaks and socio-technical abuse testing across various customers.

Required Qualifications

• Prior red-teaming experience in AI adversarial work, cybersecurity, or socio-technical probing or a strong AI background for rapid learning.
• Expertise in adversarial machine learning, including jailbreak datasets, prompt injection, RLHF/DPO attacks, and model extraction.
• Cybersecurity skills such as penetration testing, exploit development, and reverse engineering.
• Experience with socio-technical risk areas like harassment, disinformation, or abuse analysis.
• Creative probing using psychology, acting, or writing to develop unconventional adversarial methods.

Application Process

The application takes approximately 20 minutes including resume upload, a 15-minute AI interview based on your resume, and submission of a form.

Key skills/competency

adversarial testing, red teaming, AI security, cybersecurity, jailbreak, prompt injection, RLHF, penetration testing, reverse engineering, socio-technical risk