Expertise · AI infrastructure recruiting
GPU and HPC Infrastructure Recruiting
We recruit GPU cluster engineers, HPC scheduling and capacity engineers (including power-aware scheduling), hardware deployment and fleet operations engineers, telemetry engineers and storage engineers for Kubernetes and Slurm environments. We screen candidates on the size of the clusters they have run and the failures they have fixed.
Audience
We work with GPU cloud and compute platforms, AI labs, HPC centers, companies building internal training clusters, and data center operators adding compute engineers to a facilities team.
Common hiring problems
- Hardware arrives on a set date and nobody is assigned to get the cluster running.
- Job descriptions written for cloud infrastructure return candidates who have never run accelerated nodes at scale.
- Utilization is low and no one can say whether the cause is scheduling, storage, fabric or job design.
- Researchers are doing fleet operations work instead of research.
Role families
Roles and screening
GPU cluster engineering
Cluster engineer, HPC systems engineer, accelerated compute engineer
Node count and topology operated, driver and firmware lifecycle, and recovery of failed jobs on shared clusters.
Scheduling and capacity
Scheduler engineer, capacity engineer, Slurm or Kubernetes platform engineer
Queue design, fair-share policy, preemption, and power-aware and thermally constrained scheduling.
Hardware deployment and fleet operations
Hardware deployment engineer, fleet operations engineer, field systems engineer
Rack and burn-in process, RMA cycles, spares strategy and running a fleet in production.
Fleet telemetry and observability
Telemetry engineer, infrastructure observability engineer
GPU-level metrics, per-job attribution, thermal and power telemetry, and acting on what the data shows.
Storage and data path
Storage infrastructure engineer, parallel filesystem engineer
Throughput at training scale, checkpoint behavior and the interaction between storage and interconnect.
Environments
Technologies and environments
- NVIDIA GPU fleets, DGX and OEM reference systems
- Slurm, Kubernetes, Kubeflow and job schedulers
- Parallel and object storage, high-throughput checkpointing
- Power and thermal constraints at rack and row level
- Firmware, driver and image lifecycle management
Role scope
Requirement mistakes
Sort these out before outreach starts.
- Requiring a specific scheduler when the underlying skill is queue and capacity design.
- Screening for years of Kubernetes without asking whether the candidate has run GPU workloads on a busy shared cluster.
- Combining fleet operations, scheduling and platform engineering into one requisition.
- Setting the pay range from general cloud engineering salary data without checking what GPU cluster candidates are asking for.
Process
Search process
Czarina Tabayoyong agrees on the scorecard with your interview panel, builds a list of named candidates and contacts them directly. You get a written report every week. At the end, your team keeps all the research and the candidate pipeline. We use AI tools for research, and Czarina makes the recommendations herself.
Talent Intelligence Sprint
A fixed-scope research project on the candidate market before you commit to a search. Scopes are listed on the homepage.
Embedded Search Sprint
A recruiter inside your process for a defined sprint.
Site Launch or Team Build
Monthly recruiting support timed to your project milestones.
FAQ
Recruiting questions
- How do you tell a GPU cluster engineer from a cloud infrastructure engineer?
- We ask about their work. A GPU cluster engineer can describe node counts and topology they have run, how they handled a failed collective, what firmware and driver drift did to a training run, and how the scheduler behaved under contention. Cloud infrastructure engineers often haven't done any of that, even with years of Kubernetes.
- Do you recruit for Slurm environments as well as Kubernetes?
- Yes. Many training environments run both, and queue policy, capacity allocation and failure recovery carry over between them. We screen for those and treat a specific scheduler as a preference unless your environment requires it.
- What does GPU fleet operations cover?
- Physical and logical fleet health: rack and burn-in, firmware and driver lifecycle, spares and RMA cycles, node drain and return-to-service, and the telemetry that shows whether a node should be trusted with a long job.
- Can you help decide what to hire first?
- Yes. A typical order is one operations-capable engineer who can keep the cluster usable, then fabric and networking depth, then platform engineering to remove toil, then a reliability owner once you commit to availability targets.
- How do you handle compensation in this market?
- We report what candidates ask for on our calls and compare it with your approved range. Then we look at what you can change: scope, hardware access, ownership, remote policy and equity.
Network and fabric recruiting
InfiniBand and high-performance Ethernet engineering for training clusters.
Platform, SRE and observability recruiting
Platform engineering and reliability for the services that run on the cluster.
AI infrastructure recruiting
Our recruiting across compute, platform, reliability and model serving.
Next step
Bring the scorecard and the deadline. We'll tell you what the candidate market looks like and which engagement we'd recommend, including when your team can fill the role internally.