Member of Technical Staff – Integration & RL (Research Engineer)

About the role

The Integration/RL team at Cohere builds and scales machine‑learning algorithms and infrastructure for large‑language‑model post‑training, focusing on large‑scale, distributed reinforcement‑learning methods. You will help improve the global quality of the post‑training codebase by creating new research tools, optimizing algorithms, and pushing distributed RL to new performance levels. The role is remote‑friendly and can be based in Paris or anywhere within the UTC‑06:00 to UTC+01:00 time zones.

Key responsibilities

• Design and implement high‑performance, scalable software for training LLMs.
• Develop tools that accelerate research and LLM training pipelines.
• Collaborate with infrastructure, efficiency, serving, and scientific teams to build an integrated post‑training ecosystem.
• Improve training speed and performance across supervised fine‑tuning, offline preference learning, and RL regimes.
• Research, prototype, and experiment on Cohere’s cluster and data infrastructure.

Required profile

• Extremely strong software engineering abilities with a focus on clean, test‑driven code.
• Experience debugging large‑scale distributed training strategies.
• Ability to work collaboratively with engineers and scientists of varying expertise.

Required skills

• Python programming.
• ML frameworks such as JAX, PyTorch, and XLA/MLIR.
• Distributed training debugging and profiling.
• Kubernetes (bonus).
• Ray (bonus).

What we offer

• Weekly lunch stipend of $75 (or local equivalent).
• Full health, dental, and mental‑health benefits.
• Retirement plans (RRSP matching, 401k, pension).
• 100% parental‑leave top‑up for up to six months.
• Six weeks of paid vacation and travel budget for office visits.
• Home‑office stipend and co‑working benefits.