AI Safety Expert – English, Bengali

This is a fully remote position, open to applicants in United States.

📋 Description

• Engage in red teaming for conversational AI models and agents by conducting jailbreaks, prompt injections, misuse scenarios, bias exploitation, and multi-turn manipulations.

• Create human data by annotating failures, categorizing vulnerabilities, and identifying systemic risks.

• Utilize taxonomies, benchmarks, and playbooks to ensure consistent testing procedures.

• Generate reproducible reports, datasets, and attack cases for clients.

• Evaluate AI outputs related to sensitive issues such as bias, misinformation, and harmful behaviors.

• Enhance evaluation coverage and fortify customer AI systems.

⛳️ Requirements

• Proficient in both English and Bengali is essential.

• Previous experience in red teaming for AI adversarial tasks, cybersecurity, or socio-technical probing is required.

• Capable of adversarially probing AI systems and pushing them to their limits.

• Familiarity with frameworks or benchmarks for structured testing is necessary.

• Ability to communicate risks effectively to both technical and non-technical stakeholders.

• Flexibility to adapt across various projects and clients.

• Preferred skills include expertise in adversarial machine learning, cybersecurity, socio-technical risk, and innovative probing techniques.

• Must operate as an independent contractor.

• H1-B and STEM OPT candidates are not eligible.

🏝️ Benefits

• Completely remote position.

• Flexible work schedule allowing you to set your own hours.

• Weekly payments processed via Stripe or Wise.

• Project durations may vary based on needs and performance, with possibilities for extensions or early conclusions.

• Access to wellness resources and clear guidelines for sensitive projects.

• Competitive compensation.

• Opportunity to work alongside leading researchers.

• Referral bonuses of up to $90 for each successful referral, with no limit on the number of referrals (certain restrictions may apply).