AI Trainers (Hebrew Language- Native speakers)

Our Client partners with leading AI teams to improve the quality, usefulness, and reliability of general-purpose conversational AI systems. These systems are used across a wide range of everyday and professional scenarios, and their effectiveness depends on how clearly, accurately, and helpfully they respond to real user questions.

This project focuses on evaluating and improving general chat behavior in large language models (LLMs). You will assess model-generated responses across diverse topics, provide high-quality human feedback, and help ensure AI systems communicate in ways that are accurate, well-reasoned, and aligned with human expectations.

What You’ll Do

Evaluate LLM-generated responses on their ability to effectively answer user queries

Conduct fact-checking using trusted public sources and external tools

Generate high-quality human evaluation data by annotating response strengths, areas for improvement, and factual inaccuracies

Assess reasoning quality, clarity, tone, and completeness of responses

Ensure model responses align with expected conversational behavior and system guidelines

Apply consistent annotations by following clear taxonomies, benchmarks, and detailed evaluation guidelines

Who You Are

You hold a Bachelor’s degree

You are a native speaker or have ILR 5/primary fluency (C2 on the CEFR scale) in Hebrew

You have significant experience using large language models (LLMs) and understand how and why people use them

You have excellent writing skills and can clearly articulate nuanced feedback

You have strong attention to detail and consistently notice subtle issues others may overlook

You are adaptable and comfortable moving across topics, domains, and customer requirements

You have a background or experience in domains requiring structured analytical thinking (e.g., research, policy, analytics, linguistics, engineering)

You have excellent college-level mathematics skills

Nice-to-Have Specialties

Prior experience with RLHF, model evaluation, or data annotation work

Experience writing or editing high-quality written content

Experience comparing multiple outputs and making fine-grained qualitative judgments

Familiarity with evaluation rubrics, benchmarks, or quality scoring systems

What Success Looks Like

You identify factual inaccuracies, reasoning errors, and communication gaps in model responses

You produce clear, consistent, and reproducible evaluation artifacts

Your feedback leads to measurable improvements in response quality and user experience

Mercor customers trust the quality of their AI systems because your evaluations surface issues before public release