Adversarial Prompt Expert

@ Handshake

Overview

You’ll be part of a red teaming project focused on probing large language models for failure modes and harmful outputs. In this role as an Adversarial Prompt Expert, you will craft prompts and scenarios to test model guardrails, explore creative ways to bypass restrictions, and systematically document outcomes.

You will think like an adversary to uncover weaknesses and collaborate with engineers and safety researchers to improve system defenses.

Details

This is a remote and asynchronous role with flexible hours (approximately 10 to 20 hours per week). Project work includes developing domain-specific prompts and evaluating LLM responses. You will also research interesting topics with AI assistance and continuously learn new skills while contributing to the evolving world of AI.

Placement into a project depends on availability.

Qualifications

• Heavy LLM Usage – hands-on experience with multiple models.
• Prompt Engineering & Jailbreaking – skilled in crafting prompts and evasion techniques.
• Adversarial / Security Mindset – ability to think like an attacker with red teaming knowledge.
• Persistence & Creativity – willingness to try variations and think outside the box.
• Clear Documentation – systematic logging and clear communication of outcomes.
• Ethical Awareness – understanding of boundaries and sensitive content handling.

This program is open to U.S.-based doctoral students, candidates, and recent graduates with valid work or training authorization. Applicants must ensure compliance with their visa conditions.

Key skills/competency

• LLM
• prompt engineering
• adversarial
• red teaming
• jailbreaking
• security
• documentation
• creativity
• research
• ethical