AI Safety Experts — English & Odia | $20-$22/hr Remote
Overview
Mercor is looking for native English and Odia speakers to join a remote red team that stress-tests AI conversational models. You'll probe for vulnerabilities using adversarial techniques like jailbreaks and prompt injections, generating critical data that makes AI systems safer. This text-based project covers sensitive topics with clear guidelines and wellness support, allowing you to work on high-impact safety work from anywhere.
What You'll Do4
- 1Perform adversarial testing on AI conversational agents by designing and executing jailbreaks, prompt injections, misuse cases, and multi-turn manipulation to expose weaknesses.
- 2Generate high-quality human red team data by annotating model failures, classifying vulnerability types, and flagging systemic risks in AI outputs.
- 3Follow structured taxonomies, benchmarks, and testing playbooks to ensure consistent and repeatable evaluation across different scenarios.
- 4Document findings reproducibly by creating reports, datasets, and attack case studies that customers can directly use to harden their AI systems.
Requirements4
- 1Native-level fluency in English and Odia (spoken and written) is mandatory.
- 2Proven experience in AI adversarial testing, cybersecurity, or socio-technical probing — you're skilled at pushing systems to their breaking points.
- 3A structured approach to testing: you rely on frameworks, benchmarks, and playbooks rather than random experimentation.
- 4Strong communication skills to explain technical risks clearly to both technical and non-technical stakeholders, and adaptability to work across multiple projects and customers.
Who Should Apply
This role is ideal for experienced red teamers, security researchers, and adversarial AI testers who enjoy methodically uncovering hidden vulnerabilities. You should be fluent in English and Odia, comfortable working with sensitive content, and passionate about improving AI safety through high-quality red team data. Backgrounds in penetration testing, jailbreak research, or socio-technical risk analysis are a strong plus.
Salary Insight
The role pays $20.00 - $22.00 per hour on an hourly, remote contract basis.
Application Tip
Showcase specific examples of adversarial testing you've performed — such as crafting a jailbreak that exposed a model bias or documenting a multi-turn manipulation technique. Including your approach to recording and reporting findings will demonstrate the structured, reproducible mindset this role demands.
Similar open positions
Explore active roles that match your skills and interests.
Mercor
VerifiedAI Safety Experts — English & Assamese | $20-$22/hr Remote
We're assembling a remote team of AI red teamers to stress-test conversational models and uncover hidden vulnerabilities. In this role, you'll probe AI systems with adversarial inputs—jailbreaks, prompt injections, bias exploitation—and generate the high-quality human data that makes AI safer for real-world use. Fluency in both English and Assamese is essential, as you'll review sensitive outputs and document reproducible attack cases. This project is ideal for structured, curious thinkers who enjoy pushing systems to their limits.
Mercor
VerifiedAI Safety Experts — English & Bengali | $20-$22/hr Remote
Neon is seeking bilingual (English & Bengali) AI safety experts to join a remote red teaming project. In this role, you'll proactively probe conversational AI models using adversarial techniques like jailbreaks and prompt injections to uncover vulnerabilities before they reach production. Your work directly contributes to making AI systems more robust and trustworthy.
Mercor
VerifiedAI Safety Experts — English & Punjabi | $20-$22/hr Remote
We're looking for bilingual experts fluent in English and Punjabi to help stress-test and improve the safety of AI conversational models. In this remote, project-based role, you'll probe AI systems for hidden vulnerabilities—from jailbreaks to bias exploitation—and deliver actionable data that makes these models more trustworthy. This is hands-on adversarial testing with a direct impact on real-world AI safety.
Mercor
VerifiedAI Safety Red Teamer | $70-$84/hr Remote
Join a team focused on strengthening the safety of advanced AI systems by uncovering their hidden weaknesses. As an AI Safety Red Teamer, you'll design clever prompts to stress-test models, spot dangerous behaviors, and help improve how these systems handle tricky real-world scenarios. This remote contractor role lets you work alongside top researchers while earning $70–$84 per hour.
Mercor
VerifiedGeneralist - English & Odia | $15-$20/hr Remote
Join a global team of evaluators helping refine AI-generated responses in Odia. In this contract role, you’ll analyze model outputs for accuracy, clarity, and cultural relevance, then provide detailed feedback in English. Your work directly shapes the next generation of conversational AI.
Mercor
VerifiedCybersecurity Experts | $70-$90/hr Remote
Mercor is teaming up with a premier AI research lab to recruit seasoned cybersecurity professionals for a short-term, high-impact project. In this role, you'll leverage your expertise in low-level systems programming and security vulnerability classification to help train AI models in threat detection and reasoning. The engagement starts with a paid work trial and could extend into a roughly two-month project, all fully remote.