Shaping the Future of AI: The Mercor English & Malay AI Safety Expert Remote Career Opportunity
Company: Mercor · Pay: $17–$25/hr · Location: Remote — Global · Type: contractor · Languages: Malay · Posted: 2026-09-15
As artificial intelligence scales globally, the demand for multilingual experts who can audit, stress-test, and secure large language models has never been higher. Mercor is currently hiring for the remote position of AI Safety Experts — English & Malay, offering an attractive hourly compensation rate of $17 to $25 per hour. This comprehensive guide details the role requirements, day-to-day responsibilities, payment structures, and a complete application roadmap to help you secure this position.
AI Safety Experts — English & Malay
Role Summary: Join Mercor's red team to probe conversational AI models with adversarial inputs, surface systemic vulnerabilities, and generate critical safety data for global tech enterprises.
Understanding the Role: What is AI Red Teaming?
At Mercor, safety is achieved by proactive testing. Artificial intelligence models must undergo rigorous examination by human experts before deployment to ensure they do not produce biased, harmful, or misleading outputs. As an AI Safety Expert specializing in English and Malay, your core objective is to act as an ethical adversary—pushing conversational AI models to their absolute breaking points.
You will explore jailbreak prompts, prompt injections, multi-turn manipulation tactics, and bias exploitation cases. Your work directly safeguards future users by building structured datasets and attack reports that developers use to patch vulnerabilities.
Bilingual Insight / Wawasan Dwiperingkat (English & Malay)
To give you a clearer perspective of the cross-lingual requirements, here is an overview of the duties and qualifications presented side-by-side in English and Malay:
English Version
What You’ll Do:
- Red team conversational AI models using jailbreaks and prompt injections.
- Classify vulnerabilities and generate structured human training data.
- Document attack cases clearly for technical and non-technical stakeholders.
What We Look For:
- Native fluency in both English and Malay.
- Curious and adversarial mindset.
- Familiarity with structured testing frameworks.
Versi Bahasa Melayu
Tugas Anda:
- Melakukan ujian "red team" ke atas model AI perbualan menggunakan teknik jailbreak dan suntikan gesaan.
- Mengelaskan kelemahan dan menjana data latihan manusia berstruktur.
- Merdokumenkan kes serangan dengan jelas untuk pihak berkepentingan teknikal dan bukan teknikal.
Ciri-Ciri Calon Yang Dicari:
- Fasih sepenuhnya dalam Bahasa Inggeris dan Bahasa Melayu.
- Minda yang ingin tahu dan berminat menguji sistem secara mencabar.
- Kebiasaan dengan kerangka kerja pengujian berstruktur.
Comprehensive Job Breakdown & Key Responsibilities
The position is structured around several critical pillars designed to enhance the safety and resilience of large language models:
- Adversarial Probing: Invent creative scenarios, linguistic traps, and edge cases in both English and Malay to test model safety limits.
- Data Annotation: Label system failures, categorize toxicity or misinformation vectors, and ensure strict adherence to safety frameworks.
- Reproducible Documentation: Produce precise reports and testing datasets that outline how specific jailbreaks or behavioral anomalies occurred.
- Cross-Cultural Nuance Application: Leverage your deep native understanding of cultural expressions, idioms, and regional variations in Malay and English to evaluate context sensitivity.
💡 Note on Wellbeing & Support: Reviewing AI outputs that touch upon sensitive topics (such as bias or safety edge cases) is managed with clear guidelines, advance topic warnings, and supportive wellness resources.
Ideal Candidate Profile & Qualifications
Mercor looks for sharp, adaptable, and structured thinkers who can approach technology with a healthy dose of skepticism:
- Linguistic Mastery: Native or near-native fluency in both English and Malay is an absolute prerequisite.
- Analytical Curiosity: An innate drive to uncover hidden system flaws, bypass standard parameters, and think outside the box.
- Structured Methodology: Ability to apply systematic benchmarks, taxonomies, and playbooks rather than random hacking.
- Bonus Specialties: Backgrounds in cybersecurity (penetration testing), socio-technical risk analysis, adversarial machine learning (RLHF/DPO attacks), or creative writing/psychology are highly valued.
Compensation, Contract Terms, and Earning Potential
Contractors partnering with Mercor enjoy industry-competitive pay scales, weekly disbursements, and complete remote autonomy:
- Hourly Wage: Earn between $17 and $25 per hour based on project specifications and your expert contributions.
- Flexible Schedule: Complete your assignments on your own schedule from anywhere with reliable internet access.
- Weekly Payouts: Reliable weekly compensation processed seamlessly through Stripe or Wise.
- Referral Bonuses: Mercor features a referral program where you can earn bonuses by referring successful candidates.
Step-by-Step Application Roadmap
To maximize your chances of acceptance into Mercor's elite AI safety pools, follow these crucial steps:
- Prepare Your Resume: Highlight any relevant experience in translation, linguistic review, data annotation, cybersecurity, or academic research.
- Demonstrate Bilingual Excellence: Clearly show your proficiency in both English and Malay within your profile and application screening questions.
- Complete Screening Promptly: Mercor utilizes a streamlined interview process. Respond to invitations and complete assessments without delay to maintain early applicant advantage.
Take the next step in your career by submitting your complete application today!
Responsibilities
- Adversarial Probing: Invent creative scenarios, linguistic traps, and edge cases in both English and Malay to test model safety limits.
- Data Annotation: Label system failures, categorize toxicity or misinformation vectors, and ensure strict adherence to safety frameworks.
- Reproducible Documentation: Produce precise reports and testing datasets that outline how specific jailbreaks or behavioral anomalies occurred.
- Cross-Cultural Nuance Application: Leverage your deep native understanding of cultural expressions, idioms, and regional variations in Malay and English to evaluate context sensitivity.
Qualifications
- Linguistic Mastery: Native or near-native fluency in both English and Malay is an absolute prerequisite.
- Analytical Curiosity: An innate drive to uncover hidden system flaws, bypass standard parameters, and think outside the box.
- Structured Methodology: Ability to apply systematic benchmarks, taxonomies, and playbooks rather than random hacking.
- Bonus Specialties: Backgrounds in cybersecurity (penetration testing), socio-technical risk analysis, adversarial machine learning (RLHF/DPO attacks), or creative writing/psychology are highly valued.