职位描述

Role Overview

This contract position seeks a Cybersecurity Consultant specializing as an LLM Red-Teamer. The consultant will perform adversarial testing to detect vulnerabilities and enhance the resilience of large language models. The work involves designing and executing scenarios to challenge model behavior under edge cases and potentially malicious inputs.

Key Responsibilities

Create and carry out red-team strategies aimed at identifying flaws in LLM outputs such as bias, toxicity, and factual errors.
Build both automated and manual testing frameworks that emulate adversarial attacks on AI models.
Prepare detailed documentation of findings, identifying risks and suggesting mitigation approaches.
Work collaboratively with AI safety researchers to update testing protocols based on new threat developments.
Keep abreast of innovations in AI security, including prompt injection and jailbreak attack methods.

Required Qualifications and Skills

Hands-on experience in adversarial testing of large language models, involving prompt engineering and attack simulation.
Strong proficiency in Python programming and familiarity with AI/ML libraries such as Hugging Face or PyTorch.
Understanding of cybersecurity principles, specifically those that apply to AI systems.
Capability to analyze model outputs for issues like bias, toxicity, and hallucinations.
Excellent written communication skills for comprehensive vulnerability reporting and recommendation writing.
Awareness of ethical AI standards and responsible AI development practices.

Additional Information

This opportunity allows professionals to collaborate with a leading global technology and internet company, influencing safer AI system deployment by exposing critical vulnerabilities before public release.

Equal Opportunity Employer Statement

Hiring decisions are based purely on skills and qualifications. All eligible candidates, regardless of background or prior experiences, will be fairly considered.

Cybersecurity Consultant (Remote)