Note: The job is a remote job and is open to candidates in USA. Shadeform is a GPU cloud marketplace that aggregates compute across multiple cloud providers and data centers. They are seeking a Machine Learning Engineer to build an optimization and orchestration layer for managing ML workloads across a heterogeneous fleet of hardware.
Responsibilities
- Build the optimization and orchestration logic that places and tunes ML workloads across a heterogeneous, multi-provider fleet
- Benchmark GPUs, interconnects, and driver stacks to understand real per-node performance, and feed that back into how the platform makes decisions
- Build the tooling, images, and reference stacks that take someone from a fresh instance to a running job without a day of setup
- Dig into performance and reliability problems that cross the line between the workload and the underlying hardware
- Turn what you learn into things that scale: defaults, playbooks, and platform improvements
- Take part in a weekly on-call rotation (24/7 coverage shared across the team)
Skills
- Solid experience getting ML workloads running in production, not just notebooks or research code
- A real understanding of GPUs and how ML workloads use them: memory, throughput, and where things actually bottleneck
- Comfort down at the systems and Linux level. You're not lost when a problem turns out to be drivers or networking rather than the model
- The ability to own an ambiguous problem end to end and build a process where none exists yet
- Distributed training or large-scale inference in production
- Familiarity with NVIDIA driver software, interconnects (InfiniBand/RoCE), or cluster networking
Benefits
- A remote-first team working across the US and EU.
- Comprehensive health benefits and flexible time off.
Company Overview