AI Safety Expert for English and Assamese

Conduct adversarial testing of conversational AI in English and Assamese to find and document safety failures, generate reproducible attack cases, and produce high-quality human data that helps customers harden their systems. The work is text-based and will include reviewing outputs on sensitive topics such as bias, misinformation, and harmful behaviors. Participation in higher-sensitivity tasks is optional, supported by clear guidelines and wellness resources, and any sensitive topics will be communicated to you before exposure.

• Red team conversational AI models and agents, including jailbreaks, prompt injection, misuse cases, bias exploitation, and multi-turn manipulation.
• Generate high-quality human data by annotating failures, classifying vulnerabilities, and flagging systemic risks.
• Follow structured taxonomies, benchmarks, and playbooks to keep testing consistent and reproducible.
• Document findings clearly, producing reports, datasets, and attack cases that customers can act on.

• Native fluency in English and Assamese, both written and spoken, is required.
• Prior red teaming or adversarial experience, such as AI adversarial work, cybersecurity, or socio-technical probing.
• Ability to work methodically with frameworks and benchmarks rather than ad hoc approaches.
• Strong communication skills, able to explain risks to technical and non-technical stakeholders.
• Curiosity and an adversarial mindset, with adaptability to move across projects and customers.

• Adversarial machine learning, including jailbreak datasets, prompt injection, RLHF or DPO attack techniques, or model extraction.
• Cybersecurity experience, such as penetration testing, exploit development, or reverse engineering.
• Socio-technical risk analysis, including harassment, disinformation probing, abuse analysis, or conversational AI testing.
• Creative probing skills, including psychology, acting, or writing for unconventional adversarial approaches.

• Uncovering vulnerabilities that automated tests miss.
• Delivering reproducible artifacts that help customers strengthen their AI systems.
• Expanding evaluation coverage so fewer surprises occur in production.
• Helping customers trust their AI because it has been probed like an adversary.

• Location: Remote.
• Employment type: Hourly.
• All work is text-based.
• Higher-sensitivity projects are optional and will be accompanied by guidelines and wellness resources, and topics will be communicated before exposure.

• Pay rate: 20 - 22 hourly.

• Required: Native fluency in both English and Assamese.