Software Engineer, ML Workload Platform

Cohere is building the infrastructure and tools to train, evaluate, and serve its foundation models. This role involves owning the platform that schedules and runs training, evaluation, and reinforcement-learning jobs on Kubernetes, building networking and runtime isolation, working with data infrastructure on storage, troubleshooting production failures, and building automation and observability. The team works with AI researchers and ML engineers to accelerate the models behind Cohere's North.

Experience running a job platform, batch system, or Kubernetes controllers in production with scheduling, queues, and multi-tenant fairness. Distributed systems judgment including retries and partial failure. Worked on ML training workflows, RL sandboxes, container networking, or checkpoint storage at scale. Written software in Python or Go. Experience working with researchers or ML engineers. Ability to find bottlenecks, propose fixes, and drive in a fast-moving environment.