Mercor
Mercor
51 – 200 Employees
Mercor’s available profile does not include enough verified information about its products, services, customers, sector, or hiring focus to support an accurate company overview. More source material is needed before describing the business or its opportunities.

AI Safety Expert – English and Marathi (Remote)

Mercor is seeking English- and Marathi-speaking AI safety experts to red-team conversational models and agents. The work involves identifying vulnerabilities, evaluating sensitive outputs, and creating reproducible safety data for AI labs and customers.

Description

  • Conduct red-team testing of conversational AI models and agents using jailbreaks, prompt injections, misuse scenarios, bias exploitation, and multi-turn manipulation
  • Create human-evaluation data by annotating failures, categorizing vulnerabilities, and identifying systemic risks
  • Use established taxonomies, benchmarks, and playbooks to support consistent evaluations
  • Develop reproducible reports, datasets, and attack cases for customers
  • Assess AI outputs related to bias, misinformation, harmful behavior, and other sensitive topics
  • Identify vulnerabilities that automated testing may overlook
  • Broaden evaluation coverage to help minimize unexpected production issues
  • Support customers in improving the safety, robustness, and trustworthiness of their AI systems

Requirements

  • Native fluency in both English and Marathi
  • Strong judgment when evaluating language and content, including whether AI responses are accurate, complete, and appropriate
  • Ability to detect subtle errors, inconsistencies, and omissions
  • Ability to follow guidelines and quality standards consistently
  • Ability to communicate reasoning clearly to technical and non-technical audiences
  • Adaptability across different projects, task types, and customers
  • Ability to work as an independent contractor
  • Must not require H-1B or STEM OPT support
  • Preferred: adversarial machine learning experience with jailbreak datasets, prompt injection, RLHF or DPO attacks, or model extraction
  • Preferred: cybersecurity experience in penetration testing, exploit development, or reverse engineering
  • Preferred: socio-technical risk experience in harassment or disinformation probing, abuse analysis, or conversational AI testing
  • Preferred: psychology, acting, or writing experience that supports unconventional adversarial thinking

Benefits

  • Fully remote work that can be completed on your own schedule
  • Weekly payments through Stripe or Wise
  • Optional participation in higher-sensitivity projects
  • Clear guidelines and wellness resources for work involving sensitive content
  • Competitive pay
  • Opportunity to collaborate with leading researchers
  • Referral payments of up to $90 per successful referral, subject to limits

Related Jobs