ML Workload Platform Engineer - Remote

Cohere; United States; Remote; USD 99,000 - 142,000; Full time

Benefits offered by this job

Weekly lunch stipend

Health and dental benefits

RRSP matching and pension plans

Parental leave top-up

Enrichment benefits (conferences, stip

Education & learning stipend

6 weeks paid vacation

Travel budget for other offices

Home office stipend $500

Job summary

Cohere is seeking a Platform Engineer to own and operate the job scheduling and runtime platform for training, evaluation, and reinforcement-learning workloads. You will work with researchers and ML engineers to ensure stability, scalability, and observability of the platform.

You will build networking and isolation for workloads, partner with data storage teams for checkpoints, and drive automation and reliability in a fast-moving environment.

Qualifications

• Experience running a job platform, batch system, or Kubernetes controllers in production.

Responsibilities

• Own the platform that schedules and runs training, evaluation, and reinforcement-learning jobs on Kubernetes.
• Build the networking and runtime isolation those workloads need.
• Collaborate with data infrastructure on storage for checkpoints and artifacts.
• Troubleshoot production failures and improve operability.
• Build automation and observability to keep the platform maintainable.

Job description

Cohere is seeking a Platform Engineer to own and operate the job scheduling and runtime platform for training, evaluation, and reinforcement-learning workloads. You will work with researchers and ML engineers to ensure stability, scalability, and observability of the platform.

You will build networking and isolation for workloads, partner with data storage teams for checkpoints, and drive automation and reliability in a fast-moving environment.