This position is listed on behalf of a partner company, who manages all applications and next steps. Our partner is looking for an AI Safety Expert — English & Odia based in India.
This is a remote opportunity for experts who want to help make conversational AI systems safer, more reliable, and more resilient. You will take part in adversarial testing designed to uncover vulnerabilities that automated evaluations may overlook. Your work will involve probing AI models and agents for jailbreaks, prompt injection, misuse scenarios, bias, and multi-turn manipulation. You will transform your findings into structured annotations, attack cases, datasets, and actionable reports. The role offers exposure to cutting-edge AI safety projects and the prospect to contribute directly to the development of more trustworthy AI systems. Higher-sensitivity projects are optional and supported by clear guidelines and wellness resources.
n

Accountabilities:
Red-team conversational AI models and agents by testing jailbreaks, prompt injections, misuse scenarios, bias exploitation,



and multi-turn manipulation.
Identify vulnerabilities, failure modes, systemic risks, and safety gaps that automated testing may not detect.
Generate high-quality human data by annotating model failures, classifying vulnerabilities, and documenting adversarial examples.
Apply established taxonomies, benchmarks, testing frameworks, and playbooks to ensure consistent and reproducible evaluations.
Create structured attack cases, datasets, evaluation artifacts, and reports that can be used to improve AI safety and model performance.
Clearly communicate technical and behavioral risks to both technical and non-technical stakeholders.
Adapt testing approaches across different AI safety projects, models, use cases, and customer requirements.
Contribute specialized expertise in areas such as adversarial machine learning, cybersecurity, socio-technical risk, abuse analysis, psychology, creative writing, or unconventio

📌 AI Safety Experts — English (India)
🏢 Jobgether
📍 India