Member of Technical Staff - Post-Training (Reinforcement Learning & Generalization)

Europe, Remote. Posted a month ago

The role

You will train large language models with reinforcement learning and develop methods to help them generalize. You will design training recipes, environments, and evaluations to improve production models.

What you will do

Design, run, and analyse large-scale RL post-training experiments
Research and develop methods that improve out-of-distribution generalization
Build held-out evaluations to measure generalization and catch reward hacking
Diagnose training dynamics and failure modes
Turn successful research into robust, reproducible training recipes for production models

Required qualifications

Strong research and software engineering skills
Hands-on experience training LLMs
Experience applying reinforcement learning to LLMs (e.g., PPO, GRPO, RLHF, RLVR)
Track record of work on generalization, robustness, or distribution shift through publications or shipped models
Rigorous experimental practice
Excellent communication skills
Deep understanding of data quality principles and best practices

Nice to have

Experience with distributed training (e.g., JAX, PyTorch, FSDP, Megatron)

Benefits

Weekly lunch stipend of $75/£75 or equivalent
Full health and dental benefits
Separate budget for mental health
RRSP matching, 401K, Pension Scheme
100% Parental Leave top-up for up to 6 months
Annual enrichment benefits for arts, culture, fitness, wellness, quality time, and workspace improvement
Education and learning stipend for conferences, courses, and coaching
6 weeks of paid vacation
Budget for traveling to other offices if remote
Annual company offsite
Daily lunch program and snacks for those in office
$500 home office stipend

Before you apply

May use AI-enabled tools to screen and assess applicants

About the company

Cohere is a security-first enterprise AI company that builds foundation AI models and end-to-end products. It is a global technology company headquartered in Toronto.