Role overview

Cohere is an enterprise AI company that builds foundation models and products for real-world business applications. The Data Infrastructure team is responsible for designing and operating the storage and data movement systems that support model training at scale, handling petabytes of data across multiple GPU clusters.

Key responsibilities

Design and build distributed storage systems that feed model training and evaluation workloads.
Operate these systems on Kubernetes clusters while managing petabyte-scale data.
Collaborate with researchers and training infrastructure teams to understand data access patterns and translate them into specific throughput, latency, and durability requirements.
Address networking, I/O, and consistency challenges involved in moving large datasets and model checkpoints across regions and storage backends.
Optimize for GPU utilization and time-to-insight as primary success metrics.

Requirements

Strong foundational knowledge of storage systems, including replication, consistency models, caching strategies, and data lifecycle management.
Strong coding ability in Python and Go, with willingness to learn either language if you have experience in only one.
Hands-on experience running stateful systems on Kubernetes, including Persistent Volumes, CSI drivers, and StatefulSets.
Practical experience with cloud object storage such as S3 and POSIX-style filesystems.
Experience with parallel or HPC filesystems like Weka, VAST, or Lustre is a bonus.
Familiarity with data-loading and checkpointing patterns in large-scale model training is advantageous.

What makes this role worth considering

You would be building a critical system from the ground up at a scale few teams have tackled, directly enabling frontier AI model training. Cohere offers comprehensive benefits including six weeks of paid vacation, full health and dental coverage, parental leave top-up for six months, education stipends, and home office setup support. The company maintains a remote-friendly culture with offices across Toronto, London, New York City, San Francisco, Montreal, Paris, Berlin, and Seoul.