Staff Engineer, LLM Post-Training & RL Generalization

Cohere; United States; Remote; USD 180,000 - 240,000; Full time

Benefits offered by this job

Lunch stipend

Health and dental benefits

RRSP matching / Pension

Parental leave

Enrichment benefits

Job summary

Cohere is hiring for a Member of Technical Staff in Post-Training to advance large language models through reinforcement learning and robust generalization techniques. You will design training recipes, environments, and evaluations, collaborating with data infrastructure, evaluation, and pre-training teams to push production-ready models.

The role emphasizes strong research and software engineering, with experience in RL for LLMs, distributed training, and data quality, while enabling scalable,

Qualifications

• Strong research and software engineering skills.
• Experience applying reinforcement learning to LLMs at scale.
• Track record of work on generalization or robustness.
• Rigorous experimental practice with ablations and controls.
• Experience with distributed training frameworks.
• Excellent communication skills.
• Deep understanding of data quality and dataset implications.

Responsibilities

• Design, run, and analyse large-scale RL post-training experiments (e.g., RLHF).
• Research methods to improve out-of-distribution generalization.
• Build held-out evaluations for generalization and to detect issues.
• Diagnose training dynamics and failure modes in models.
• Turn research into reproducible training recipes for production models.

Skills

LLM training

Reinforcement learning

Generalization robustness

Experimental design

Distributed training

Communication

Data quality assessment

Tools

JAX

PyTorch

FSDP

Megatron

Job description

Cohere is hiring for a Member of Technical Staff in Post-Training to advance large language models through reinforcement learning and robust generalization techniques. You will design training recipes, environments, and evaluations, collaborating with data infrastructure, evaluation, and pre-training teams to push production-ready models.

The role emphasizes strong research and software engineering, with experience in RL for LLMs, distributed training, and data quality, while enabling scalable,