Content Safety Analyst - AI Red Teaming (English & Bengali)

Role Overview

We are seeking vigilant AI Safety Analysts fluent in English and Bengali to join our remote contract team. This position involves rigorously testing conversational AI systems to uncover vulnerabilities and improve their resilience against adversarial attacks.

Key Responsibilities

• Perform adversarial testing on AI models by conducting techniques such as jailbreaks, prompt injections, and exploiting biases.
• Produce detailed, high-quality annotations of AI failures, classify weaknesses, and highlight systemic risks.
• Utilize established methodologies, including taxonomies, benchmarks, and playbooks, to ensure thorough and standardized testing.
• Prepare reproducible documentation including comprehensive reports, datasets, and examples of attacks to assist clients in addressing issues.

Candidate Requirements

• Experience in red teaming AI systems, cybersecurity, or sociotechnical probing with an adversarial focus.
• A mindset characterized by curiosity and a proactive approach to challenge systems to their limits.
• Methodical and framework-driven testing methods rather than ad hoc techniques.
• Excellent communication abilities to clearly convey risks and findings to both technical and non-technical audiences.
• Flexibility and adaptability in working across diverse projects and client environments.

Compensation and Commitment

This role offers contract-based compensation between $20 and $22 per hour. Candidates are expected to commit between 10 to 40 hours per week, working remotely from India.

Application Process

• Submit your application through the provided application mechanism.
• Your qualifications will be evaluated to ensure alignment with role requirements.
• Qualified candidates will be contacted with further instructions to proceed.