Senior Systems Engineer, Compute

Crusoe Crusoe · Data AI · San Francisco, CA - US · Cloud Engineering

Crusoe is an AI infrastructure company building and operating compute platforms for AI workloads. This role focuses on developing and scaling Linux applications for their virtualization stack, integrating with AI hardware, optimizing performance for AI/ML workloads, and troubleshooting complex system issues. The engineer will work on compute application development, AI hardware integration, kernel/hypervisor integration, performance analysis, and system-level troubleshooting.

What you'd actually do

  1. Design highly reliable and performant Linux applications used to manage our virtualization stack across thousands of AI compute servers in multiple global datacenters.
  2. Integrate Crusoe applications with a wide variety of hardware and software AI chip-vendor stacks. Build solutions to optimize and monitor virtualized hardware (GPUs, Infiniband/ROCe NICs, Ephemeral Storage, etc.) in cutting-edge AI/HPC environments.
  3. Work side by side with our Linux Kernel and Hypervisor teams to ensure our Crusoe applications are seamlessly integrated with a variety of kernels and hypervisors.
  4. Analyze and enhance the performance of the entire virtualization stack, from the hypervisor to the virtualized guest OS, with a specific focus on optimizing AI/ML workloads. This includes profiling, bottleneck identification, and implementing low-level optimizations.
  5. Diagnose and resolve complex system issues across our virtualization stack (drivers, kernel, hypervisor, guest OS, and crusoe applications). Work closely with kernel and hypervisor teams to debug and resolve integration challenges.

Skills

Required

  • Linux kernel
  • virtualization
  • hardware tuning
  • distributed systems
  • object oriented programming
  • low-level systems programming
  • Linux Systems Familiarity
  • Hardware Integration
  • Systems Design
  • Software Architecture
  • Excellent Communication Skills
  • Rapid and Agile Learner
  • Virtualization Concepts
  • CI/CD and Validation

Nice to have

  • virtualization specifically for AI/ML workloads
  • GPU virtualization
  • debugging or contributing to kernel or hypervisor code
  • configuring thousands of live compute nodes in a bare-metal production environment

What the JD emphasized

  • critical to this role
  • must
  • thousands of AI compute servers
  • cutting-edge AI/HPC environments
  • optimizing AI/ML workloads
  • complex system issues

Other signals

  • AI compute
  • virtualized AI-platforms
  • AI/HPC environments
  • optimizing AI/ML workloads