Safety Research Internship (Spring/Summer 2026)

About the Role

This Safety Research Internship at Cohere offers the opportunity to collaborate with Cohere's Modelling Safety team on implementing novel research ideas grounded in real usage and production models. The team is broadly interested in areas such as fairness, safety (especially multilingual, dialect, and cultural contexts), robustness, generalization, interpretability, safety for agents with complex read/write actions, and safety for codegen. Interns are encouraged to propose their own topics. The goal of the internship is to publish a paper in a top venue and contribute to open science, with many projects also influencing production systems. Cohere's mission is to scale intelligence to serve humanity through frontier models for developers and enterprises. The goal of an internship will be to publish a paper in a top venue and make a contribution to open science. Many internship projects do also make it into our production systems and influence our model development, though this is not a requirement, just a bonus. Your project details and topic will be collaboratively designed between you and the team.

Responsibilities

• Conduct cutting-edge machine learning research, training and evaluating production large language models.
• Focus on research projects aimed at making models better understood, safer, more reliable, more inclusive, and more beneficial for the world
• Disseminate your research results through the production of publications, datasets, and code.
• Contribute to research initiatives that have practical applications in Cohere’s product development.

Requirements

• Are currently pursuing, or in the process of obtaining, a PhD in Machine Learning, NLP, Artificial Intelligence, or a related discipline. We will also consider exceptional non-PhD candidates.
• Are eligible for work authorization in the country of employment at the time of hire and maintain ongoing work authorization throughout the internship period.
• Have experience using large-scale distributed training strategies, data annotation and evaluation pipelines, or implementing state of the art ML models.
• Are familiar with autoregressive sequence models, such as Transformers.
• Have strong communication and problem-solving skills with the ability to convey complex research findings clearly and succinctly.
• Have knowledge, or are knowledgeable, of programming languages such as Python, C, C++, Lua, or related languages.
• Have knowledge of related ML frameworks such as JAX, Pytorch and Tensorflow.
• Have previous experience in building systems based on machine learning and deep learning techniques.
• Demonstrate passion for applied NLP models and products.

Qualifications

• Demonstrated expertise through publications in top tier venues in fields such as machine learning, NLP, artificial intelligence, computer vision, optimization, computer science, statistics, applied mathematics, or data science.
• Proven ability to tackle analytical problems using quantitative methodologies.
• Proficiency in handling and analysing complex, high-dimensional data from various sources.
• Experience in applying theoretical and empirical research to real-world problem-solving.