Senior Research Scientist, Model Evaluation

About the Role

Cohere's mission is to scale intelligence to serve humanity by training and deploying frontier models for developers and enterprises building AI systems for content generation, semantic search, RAG, and agents. The company culture emphasizes individual contributions to model capabilities and customer value, with a focus on hard work and fast execution. Composed of passionate researchers, engineers, and designers, Cohere believes in the importance of diverse perspectives for building great products. Evaluation is paramount to advancing intelligence, especially as models achieve superhuman capabilities in many real-world use cases. This role focuses on developing new evaluation techniques that accurately reflect current model capabilities and define future model capabilities. As a Senior Research Scientist in Model Evaluation, you will be responsible for creating next-generation evaluation methods and infrastructure to measure LLM progress, pushing the boundaries of what these models can achieve.

Responsibilities

• Create ambitious new evaluation benchmarks that push the limits of what our models can accomplish.
• Work on highly cross-functional teams to translate model feedback into trustworthy, repeatable evaluations.
• Conduct research to advance the state-of-the-art in LLM evaluation methods, including training LLM judges; refining LLM-based data synthesis pipelines; and improving evaluation efficiency.
• Build scalable and reusable tools for digging into model performance.

Requirements

• You enjoy rapidly building prototypes that demonstrate the boundaries of what LLMs are capable of, and you have developed resources to measure those capabilities.
• You have spent dozens of hours reviewing complex data and LLM outputs to ensure high data quality.
• You are obsessive about rigorously measuring AI capabilities, and also about making sure your measurements actually align with the capabilities you care about.
• You have strong software engineering skills.

Qualifications

If some of the above doesn’t line up perfectly with your experience, we still encourage you to apply!