AI Red Teamer: LLM Safety & Stress-Testing Generalist

Handshake; Seattle (WA); On-site; USD 83,000 - 138,000; Full time

Job summary

Handshake AI seeks an AI Red Teamer (LLM Generalist) in Seattle, contract-based, 40 hours per week, to stress-test large language models and identify guardrail weaknesses.

You will craft adversarial prompts, explore edge cases, and document findings to strengthen model defenses across risk domains. Candidates should thrive in ethical, high-challenge environments and stay current on jailbreak methods.

Qualifications

• Proven hands-on experience with multiple LLMs (ChatGPT, Claude, Gemini, etc.).
• Ability to craft adversarial prompts and apply jailbreak/evasion concepts.
• Strong written communication and ethical judgment under pressure.

Responsibilities

• Craft creative prompts and multi-turn scenarios to test guardrails.
• Discover ways around safety filters using jailbreak or prompt-injection techniques.
• Explore edge cases to provoke disallowed outputs.
• Evaluate model responses with structured harm taxonomies.
• Document experiments: steps, rationale, and findings.
• Review prompts and refine adversarial prompts from team members.
• Contribute to harm taxonomy and calibration exercises.
• Collaborate with engineers and researchers to share findings.

Skills

LLM experience

Adversarial prompts

Creative problem-solving

Written communication

Ethical judgment

Self-directed / collaborative

Curiosity & persistence

Tools

LLM APIs

Evaluation tooling

Job description

Handshake AI seeks an AI Red Teamer (LLM Generalist) in Seattle, contract-based, 40 hours per week, to stress-test large language models and identify guardrail weaknesses.

You will craft adversarial prompts, explore edge cases, and document findings to strengthen model defenses across risk domains. Candidates should thrive in ethical, high-challenge environments and stay current on jailbreak methods.