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

Remote - United States, Canada, European Union$156k - $156k2w ago

Remote Staff EMEA Mental Health Artificial Intelligence Training Specialist JAX Data Quality Cody

Requirements

• Strong research and software engineering skills, with hands-on experience training LLMs. • Experience applying reinforcement learning to LLMs (e.g., PPO, GRPO, RLHF, RLVR), ideally at scale. • Track record of work on generalization, robustness, or distribution shift, through publications or shipped models. • Rigorous experimental practice: clean ablations, careful controls, and sound conclusions from noisy results. • Experience with distributed training (e.g., JAX, PyTorch, FSDP, Megatron) is a plus. • Deep understanding of data quality principles and best practices: demonstrates willingness to dive deep into data while possessing strong intuition for assessing the quality of datasets and anticipating how different data characteristics will impact model training outcomes. • Working Location • This role can be based remotely or from one of our office locations listed on the job description - there is no minimum in-office qualification requirement. We care most about hiring exceptional people regardless of locations, though please check the location listed on the posting for guidance around the core time zone or working hours alignment expected for the role. • FULL-TIME EMPLOYEES AT COHERE ENJOY THESE PERKS: • A weekly lunch stipend of $75/£75 or equivalent in your local currency for lunch. • Full health and dental benefits, including a separate budget for mental health. • RRSP matching, 401K, Pension Scheme. • 100% Parental Leave top-up for up to 6 months, for either parent.

Responsibilities

• Design, run, and analyse large-scale RL post-training experiments (e.g., RLHF, RL from verifiable rewards). • Research and develop methods that improve out-of-distribution generalization, such as reward design, environment and task diversity, curricula, and regularization. • Build held-out evaluations that measure generalization to unseen tasks, domains, and formats, and that catch reward hacking and benchmark overfitting. • Diagnose training dynamics and failure modes, including instability, reward over-optimization, and capability regressions. • Turn successful research into robust, reproducible training recipes for production models.

Benefits

• Arts & culture, fitness/wellness, quality time, and a workspace improvement credit. • Education & learning stipend for conferences, courses, and coaching. • 6 weeks of paid vacation (30 working days!) • Budget for traveling to other offices if you are remote, plus an annual company offsite. • HOW AND WHERE WE WORK: • Cohere is remote-friendly, but we also have offices in Toronto, London, New York City, San Francisco, Montreal, Paris, Berlin and Seoul with more opening soon. • For those in the office: a daily lunch program, plenty of snacks, and regular community and social events. • For those not near an office: a co-working benefit so you can work alongside others in your city. • Everyone receives a $500 home office stipend to set up your workspace properly. • If any of the above doesn’t line up exactly with your experience, we still encourage you to apply. • We strive to create an inclusive work environment for all; we welcome applicants from all backgrounds and are committed to providing equal opportunities. Should you require any accommodations during the recruitment process, please submit an Accommodations Request Form https://docs.google.com/forms/d/12a6IrLdF3kI2nonKSr4tiFuz18rLQbaeYV-JM9L4o9Q/edit, and we will work together to meet your needs. • We may use AI-enabled tools to screen and assess applicants against the criteria for this position. This helps our recruiters identify potentially qualified candidates, but it doesn't limit the applications our recruiters may review or consider.