Engineering Manager, GPU Infrastructure
Cohere · United States
About this role
**Engineering Manager, GPU Infrastructure** **About Cohere** Leading security-first enterprise AI company building cutting-edge foundation models and end-to-end products for real-world business problems. **The Team** The GPU Clusters team builds and operates superclusters that train Cohere's frontier models, sitting at the intersection of hardware, distributed systems, and AI research. **Your Role** Lead a team of GPU infrastructure engineers, set technical direction, and help scale a rapidly growing compute footprint. **Key Responsibilities** • Hire, mentor, and grow a team of GPU infrastructure engineers • Own the technical roadmap for fleet deployment, operations, and scaling of Kubernetes clusters • Partner with researchers and ML engineers on training/inference stack optimization • Work cross-functionally on planning, cost, compliance, and shared dependencies • Drive operational excellence: observability, automation, cost optimization, and vendor relationships **What We're Looking For** • Experience managing engineering or SRE teams with focus on mentorship and growth • Background running large Kubernetes compute fleets in production (multi-cloud) • Deep expertise in one GPU fleet layer (cluster operations, GPU networking, hardware) • Experience with cost optimization and capacity planning for GPU infrastructure • Track record partnering with researchers/ML engineers on reliability/cost/delivery tradeoffs **Compensation & Benefits** 💰 US: $240,000 - $380,000 USD | Canada: $300,000 - $435,000 CAD ✨ Weekly lunch stipend • Full health/dental • RRSP/401K matching • 100% parental leave top-up (6 months) • 6 weeks paid vacation • Education stipend • $500 home office stipend • Remote-friendly with offices globally **Locations**: Toronto, London, NYC, San Francisco, Montreal, Paris, Berlin, Seoul (remote-friendly)
Listing freshness
CronJobs last confirmed this listing 17h ago. If its source stops confirming the opening for seven days, this page is removed from active inventory.