/ TEAMS · Specialist engineering · GPU cluster infrastructure
Source engineers experienced in Accelerator clusters, distributed compute and operations. for business-critical production systems, safe integrations and measurable technical ownership.
2,000+ engineers in our active talent network · Regional recruiters
/ SYSTEM STACK
NVIDIA GPU node provisioning
NCCL networking
Kubernetes GPU operators
RDMA fabric design
GPU observability
/ AVAILABLE TO HIRE
Profiles show the experience we can source. Individual candidate availability is confirmed after enquiry.
Mid+ GPU Infrastructure Engineer
Inquire availability6+ years of experience · Serbia
GPU Infrastructure Engineer with 6 years in production systems, focused on NVIDIA GPU node provisioning and NCCL networking.
Languages: Serbian, English C1
Senior GPU Infrastructure Engineer
Available11+ years of experience · Romania
GPU Infrastructure Engineer with 11 years in production systems, focused on NCCL networking and Kubernetes GPU operators.
Languages: Romanian, English B2+
Senior GPU Infrastructure Engineer
Available14+ years of experience · Romania
GPU Infrastructure Engineer with 14 years in production systems, focused on Kubernetes GPU operators and RDMA fabric design.
Languages: Romanian, English B2+
Mid-level GPU Infrastructure Engineer
Inquire availability3+ years of experience · Czechia
GPU Infrastructure Engineer with 3 years in production systems, focused on RDMA fabric design and GPU observability.
Languages: Czech, English B2+
/ Engineering teams
Onectus uses its active engineering network and regional recruiting partners to search against concrete systems, environments and responsibilities. We evaluate practical experience rather than treating a framework mention or certification as proof of delivery. Particular availability and engagement terms are confirmed after an enquiry. For this role, practical knowledge of NVIDIA GPU node provisioning, NCCL networking, Kubernetes GPU operators helps define the engineering search.
/ Technical specializations
The right profile depends on your actual architecture, performance constraints, integration boundaries and operational responsibilities.
GPU cluster infrastructure responsibilities include nvidia gpu node provisioning, related nccl networking, clear test coverage and a reliable production handover. Examine the actual tooling and deployment environment.
GPU cluster infrastructure responsibilities include nccl networking, related kubernetes gpu operators, clear test coverage and a reliable production handover. Examine the actual tooling and deployment environment.
GPU cluster infrastructure responsibilities include kubernetes gpu operators, related rdma fabric design, clear test coverage and a reliable production handover. Examine the actual tooling and deployment environment.
GPU cluster infrastructure responsibilities include rdma fabric design, related gpu observability, clear test coverage and a reliable production handover. Examine the actual tooling and deployment environment.
GPU cluster infrastructure responsibilities include gpu observability, related workload isolation, clear test coverage and a reliable production handover. Examine the actual tooling and deployment environment.
GPU cluster infrastructure responsibilities include workload isolation, related nvidia gpu node provisioning, clear test coverage and a reliable production handover. Examine the actual tooling and deployment environment.
/ Project scenarios
Deploy a reliable AI compute cluster. Define constraints around nvidia gpu node provisioning, map dependencies and test how the proposed implementation behaves under real operating conditions.
Diagnose cross-node training stalls. Define constraints around nccl networking, map dependencies and test how the proposed implementation behaves under real operating conditions.
Improve GPU capacity planning. Define constraints around kubernetes gpu operators, map dependencies and test how the proposed implementation behaves under real operating conditions.
Protect shared accelerator workloads. Define constraints around rdma fabric design, map dependencies and test how the proposed implementation behaves under real operating conditions.
/ Practical evaluation
A senior engineering interview should surface directly relevant implementation details, debugging history and operational judgment.
Describe a production issue involving NVIDIA GPU node provisioning and what evidence led to a fix.
Which decisions around NCCL networking and Kubernetes GPU operators were you personally accountable for?
Describe a production issue involving Kubernetes GPU operators and what evidence led to a fix.
Which decisions around RDMA fabric design and GPU observability were you personally accountable for?
/ Sourcing process
Define platform versions, deployment constraints, traffic patterns, required integrations and ownership.
Search the active network and partner channels for platform-specific engineering experience.
Discuss real systems, code or configuration decisions, troubleshooting and change control.
Confirm role boundaries, location or remote access, availability, security and interviews.
Profiles, availability and engagement terms are confirmed for each request.
/ Questions
Yes. Tell us the relevant versions, operational environment and the responsibilities connected to NVIDIA GPU node provisioning. We qualify hands-on experience during each search.
These can be included when central to the project. We distinguish practical implementation work from general familiarity and ask about failure scenarios and testing.
We look at production decisions, trade-offs, incident diagnosis, code or infrastructure ownership and the ability to explain decisions without relying on a technology checklist.
Share your architecture, technical constraints, collaboration model and priorities. We can then discuss relevant sourcing channels, evaluation criteria and next steps.
/ Onectus · TEAMS
Tell us the technology, production responsibilities, experience level and working model. We'll activate the appropriate engineering sourcing channels.
