OCI AI Infrastructure Lead

DWI Consulting; Brussel Hoofdstad; On-site; EUR 120,000 - 160,000; Full time

Job summary

DWI Consulting seeks an AI Infrastructure Lead to provision and operate GPU compute for AI/ML workloads in a fully air-gapped OCI environment within an FSC-accredited facility.

The role focuses on building the foundation for data scientists to train and serve models, including RDMA-enabled networking, high-performance storage, and offline container registries.

Qualifications

• Experience provisioning OCI GPU compute and bare metal instances.
• Experience with OCI Networking and RDMA for distributed training.
• Ability to design air-gapped, offline environments with pre-loaded dependencies.
• Managing local/offline container registries in air-gapped settings.
• Experience with NVIDIA drivers and CUDA toolkit management.
• Familiarity with ML frameworks (PyTorch, TensorFlow) in infrastructure contexts.

Responsibilities

• Provision and run GPU compute for AI/ML workloads in a fully air-gapped OCI environment.
• Configure OCI cluster networking with RDMA for distributed training.
• Manage high-performance OCI object storage and block volumes.
• Operate Air-gap packaging with pre-loaded dependencies and no internet access.
• Oversee local/offline container registry and offline deployment pipelines.

Skills

OCI GPU compute

RDMA networking

Air-gap packaging

Offline container registry

NVIDIA driver management

CUDA toolkit management

PyTorch/TensorFlow familiarity

GPU monitoring

LLM infrastructure sizing

Tools

nvidia-smi

DCGM

CUDA toolkit

Job description

OCI AI Infrastructure Lead

Role: Lead AI Infrastructure Engineer (OCI GPU / Air-Gapped)
Duration: 12 months (Renewable)
Start: ASAP
Allocation: Full-time time (1 FTE)
Clearance: FSC required (facility-based) | PSC not required (NATO RESTRICTED level)

We are looking for an AI Infrastructure Lead to provision and run GPU compute for AI/ML workloads in a fully air-gapped OCI environment. The role must be performed from an FSC-accredited facility.

Objective: Create the infrastructure foundation on which data scientists can train and serve models.

What we’re looking for:

• OCI GPU compute and bare metal instance configuration and management
• OCI Cluster Networking with RDMA for distributed training
• High-performance OCI Object Storage and Block Volume
• OCI Autonomous Data Warehouse (ADW)
• Air-gap packaging, with all dependencies pre-loaded and no internet access
• Local, offline container registry management

Desirable:

• OKE (Kubernetes) for ML workload orchestration
• Experience sizing GPU infrastructure for LLM training
• Infrastructure-level familiarity with PyTorch and TensorFlow
• GPU monitoring (nvidia-smi, DCGM)
• NVIDIA driver and CUDA toolkit management