AI Red Team Specialist — English & Marathi Bilingual

About the role

We're building a global red team of bilingual language experts to stress-test AI systems before they reach the real world. This role is for sharp, detail-oriented thinkers fluent in both English and Marathi who want to help make AI models safer by actively trying to break them. You'll probe models with adversarial prompts, uncover weaknesses, and generate the structured feedback that trains safer, more robust AI behavior.

This is sensitive, high-impact work. You'll be reviewing AI-generated content that may touch on bias, misinformation, harmful instructions, and other edge-case behaviors — all in service of building safer AI systems for real-world deployment.

What you'll do

• Design and execute adversarial prompts ("red teaming") to surface unsafe, biased, or incorrect AI model outputs
• Evaluate and annotate AI responses across both English and Marathi language content
• Identify patterns of failure related to bias, misinformation, harmful behavior, and safety violations
• Document findings clearly and consistently to support model retraining and safety improvements
• Collaborate with a distributed team of reviewers and safety researchers to refine red-teaming methodologies
• Maintain strict confidentiality and handle sensitive content with professionalism and care

Requirements

• Native or near-native fluency in both English and Marathi (written and spoken)
• Strong critical thinking skills and attention to detail
• Comfort engaging with sensitive, adversarial, or potentially disturbing content in a professional manner
• Excellent written communication skills for documenting findings
• Ability to work independently in a fully remote setting
• Prior experience with content moderation, linguistics, AI evaluation, or data annotation is a plus but not required

Compensation

This role pays $20–$22 per hour, fully remote, with flexible scheduling. Multiple slots are open for qualified bilingual candidates worldwide.