Description
Handshake AI is hiring a contract AI Red Teer (LLM Generalist) in Seattle, Washington, for 40 hours per week. The role stress-tests large language models by designing adversarial prompts and multi-turn scenarios to uncover safety, bias, guardrail, hallucination, prompt-injection, and other vulnerabilities across text, image, voice, and agentic capabilities. Responsibilities include evaluating model responses, documenting experiments, refining adversarial prompts, contributing to harm taxonomies, and collaborating with engineers, data scientists, and researchers. The role requires strong LLM experience, creative adversarial problem-solving, clear written communication, and ethical judgment, with familiarity with Python, LLM APIs, evaluation tooling, structured annotation, and prior trust-and-safety or security-research experience preferred.
