Project Harbor Red Teaming - Juno | Remote AI Safety Evaluation

OneForma is looking for native Arabic linguists based in Switzerland , Saudi Arabia or the United Arab Emirates to join an exciting remote AI research project focused on AI safety evaluation and quality assurance.

Participants will review AI-generated responses and perform structured linguistic assessments to identify harmful content, safety concerns, and policy violations. The work involves evaluating large language model (LLM) outputs, classifying potential risks, documenting findings, and validating improvements after issues have been addressed.

This is a flexible, remote opportunity for experienced Arabic language professionals who are interested in helping improve the safety, reliability, and accuracy of next-generation AI systems.

Purpose:

The purpose of this project is to improve the safety and quality of AI chatbot systems by leveraging human expertise to evaluate model responses, identify potential risks, and provide high-quality feedback that supports ongoing AI development.

Main Requirements:

• Must be a native Arabic speaker based in Switzerland , Saudi Arabia or the United Arab Emirates .
• Linguistic background or professional language experience is preferred.
• Must be familiar with Apple products and AppleCare content.
• Experience evaluating Generative AI (GenAI) models or large language model (LLM) outputs is preferred.
• Must own a personal MacBook running the latest version of macOS Tahoe .
• Device must be personal (not shared) and have administrator privileges for software installation.
• Must confirm whether you have an AppleConnect account.
• Must be available to receive immediate, ad hoc task assignments.
• Must demonstrate professionalism, attention to detail, and the ability to follow evaluation guidelines.
• Strong analytical and written communication skills are required.

Other Important Information:

• This is a fully remote project.
• Participants will review AI-generated conversations for safety, quality, and policy compliance.
• Tasks include identifying harmful or unsafe content, classifying issues by risk level, documenting findings, and validating fixes after remediation.
• The use of AI tools, automation, or speech-to-text software is strictly prohibited.
• Completion of onboarding and certification is mandatory before production work begins.
• Direct communication with the client point of contact will be required throughout the project.

Help shape the future of trustworthy AI by using your language expertise to make AI systems safer, more accurate, and more reliable for users worldwide.