Software Engineer, GPU Infrastructure

Job Summary

To support a growing AI research and systems team, the full-time Software Engineer, GPU Infrastructure will design and manage a self-serve compute platform for training and inference workloads, ensuring efficient GPU fleet operation and resource allocation, while working remotely or from various locations including San Francisco, Ireland, London, and New York City.

Key responsibilities

• Build a self-serve compute platform for inference engineers and researchers to launch training jobs without managing GPU provisioning
• Operate the GPU fleet by managing provisioning, lifecycle, and capacity integration across multiple cloud providers
• Develop scheduling and placement logic to efficiently utilize available GPU capacity while supporting both long-running training jobs and production inference services

Required qualifications

• Deep experience with Kubernetes, including custom operators and multi-cluster federation
• Proven management of GPU clusters at scale, including knowledge of NVIDIA hardware and CUDA
• Experience orchestrating compute across multiple cloud environments such as AWS, GCP, or CoreWeave
• Strong fundamentals in distributed systems, including scheduling and fault tolerance
• Proficiency in infrastructure and systems-level programming in Go, Rust, or C++