Thai & English AI Safety Evaluator

ixolabs.ai; France; Hybride; EUR 28 000 - 32 000; Plein temps

Résumé du poste

IXO is seeking a remote, bilingual evaluator to craft prompts that probe model safety in Thai. You will classify prompts, identify adversarial patterns, and provide well-reasoned explanations of concerns and failures.

A bachelor’s degree and strong Thai/English writing skills are required, with preference for candidates in Southeast Asia. Remote work is standard; assignments are scoped and scheduled in advance.

Qualifications

• Bachelor’s degree completed or in progress.
• Thai language skills at native or near-native level.
• Business-level written English.
• Strong written reasoning and careful judgment around sensitive or dual-use information.
• Trust and safety, moderation, policy review, grading or adversarial testing experience is helpful.
• Prior AI/ML experience is not required for the generalist route.
• Being based in Thailand or wider Southeast Asia is preferred; applicants elsewhere are welcome.

Responsabilités

• Write challenging prompts that expose ambiguity, unsafe behavior or culturally sensitive failures.
• Classify prompts and conversations using the supplied evaluation rules.
• Identify adversarial wording and escalation patterns, with a reasoned account of the concern.
• Document the evidence behind each decision and distinguish clear errors from uncertainty.

Connaissances

Thai (native/near-native)

English (business writing)

Strong written reasoning

Judgment around sensitive information

Adversarial testing experience helpful

Formation

Bachelor’s degree (completed or in progress)

Description du poste

Your work at IXO
Apply bilingual judgment to IXO safety evaluations in Thai. You will create controlled test material and explain how the model handles language, context and potentially sensitive information.

Responsibilities

• Write challenging prompts that expose ambiguity, unsafe behavior or culturally sensitive failures.
• Classify prompts and conversations using the supplied evaluation rules.
• Identify adversarial wording and escalation patterns, with a reasoned account of the concern.
• Document the evidence behind each decision and distinguish clear errors from uncertainty.

Qualifications

• The generalist route requires a bachelor’s degree completed or in progress, native or near-native Thai, business-level written English, strong written reasoning and careful judgment around sensitive or dual-use information.
• Trust and safety, moderation, policy review, grading or adversarial testing experience is helpful.
• Prior AI/ML experience is not required for the generalist route.
• Being based in Thailand or wider Southeast Asia is preferred; applicants elsewhere are welcome.

Deliverables and terms
Deliverables are test prompts, classified interactions, supported evaluations and clear explanations of failure patterns. Work is remote, with scope, handling rules and schedule agreed before assignments begin. The work remains within authorized evaluation conditions.

Pay is $23 • $27/hr in USD.