Remote Latin American Spanish AI Evaluator (Part-Time)

mercor; Colombia; Remote; COP 98.094.000 - 124.847.000; Teilzeit

Zusammenfassung

mercor is seeking an Internet-Native Bilingual Evaluator Expert (Spanish - Latin America) on a remote contract basis. You will evaluate AI outputs for accuracy in online communication, culture, and multilingual interactions, focusing on Latin American Spanish and English.

Responsibilities include analyzing social media content for naturalness and cultural appropriateness, developing evaluation rubrics, and testing models with real-world multilingual prompts.

Qualifikationen

• Native or near-native fluency in Latin American Spanish and professional fluency in English.
• Strong writing, analytical, and critical thinking skills.
• Deep familiarity with internet culture and online communities in Spanish-speaking Latin America.
• Experience with evaluating multilingual content and online behavior.

Aufgaben

• Evaluate AI model outputs across English and Spanish to ensure accuracy in online communication and cultural references.
• Analyse content from social media, forums, and online communities for naturalness, fluency, and cultural appropriateness.
• Identify issues with slang, memes, abbreviations, and evolving online language in Latin America.
• Develop rubrics and guidelines for evaluating bilingual and internet-native communication.
• Test language models using multilingual prompts and provide structured feedback to improve performance.
• Collaborate with project leads to maintain dataset consistency and improve evaluation frameworks.

Kenntnisse

Spanish LATAM

English

Writing

Analytical thinking

Jobbeschreibung

mercor is seeking an Internet-Native Bilingual Evaluator Expert (Spanish - Latin America) on a remote contract basis. You will evaluate AI outputs for accuracy in online communication, culture, and multilingual interactions, focusing on Latin American Spanish and English.

Responsibilities include analyzing social media content for naturalness and cultural appropriateness, developing evaluation rubrics, and testing models with real-world multilingual prompts.