Staff Engineer, LLM Post-Training & RL

Cohere; New York (NY); Teletrabalho; USD 170 000 - 210 000; Tempo integral

Vantagens oferecidas por esta oferta de emprego

Lunch stipend

Health & dental benefits

Parental leave

Enrichment benefits

Vacation days

Office travel budget

Home office stipend

Resumo da oferta

Cohere is seeking a proactive Member of Technical Staff in Post-Training to advance RL-based training for large language models. You will design experiments, curate environments, and collaborate with data and evaluation teams to push generalization beyond training distributions.

This role supports remote work with multiple office options, offering a strong research culture, health benefits, and a comprehensive enrichment program that rewards ongoing learning and collaboration.

Qualificações

• Strong research and software engineering skills with hands-on LLM training experience.
• Experience applying reinforcement learning to LLMs at scale.
• Track record on generalization or robustness in models.

Responsabilidades

• Design and run large-scale RL post-training experiments like RLHF and PPO.
• Develop methods to improve out-of-distribution generalization.
• </p>
• Build held-out evaluations for unseen tasks and domains.
• Diagnose training dynamics and failure modes in large models.
• Translate research into robust training recipes for production models.

Conhecimentos

LLM training

Reinforcement learning

Generalization

Robustness

Distributed training

Communication

Data quality

Ferramentas

JAX

PyTorch

FSDP

Megatron

Descrição da oferta de emprego

Cohere is seeking a proactive Member of Technical Staff in Post-Training to advance RL-based training for large language models. You will design experiments, curate environments, and collaborate with data and evaluation teams to push generalization beyond training distributions.

This role supports remote work with multiple office options, offering a strong research culture, health benefits, and a comprehensive enrichment program that rewards ongoing learning and collaboration.