Capabilities

Wirewalk research infrastructure capabilities

A concise, printable overview of Wirewalk HPC and NVIDIA GPU engineering, fabric and storage diagnostics, acceptance testing and research operations.

Lenny Shovsky · Wirewalk
HPC, AI and research infrastructure engineering
sales@wirewalk.com · 917-217-7975 · www.wirewalk.com

## Help researchers do more, faster I design, build and look after the systems that help people run experiments, work with larger datasets and get results sooner. Work can begin with one difficult job, a new cluster or a defined acceptance question.

Compute and GPU systems

Cluster architecture and provisioning; NVIDIA GPU software and containers; Slurm and research workload operations.

Fabric and storage

InfiniBand, RoCE and Ethernet; GPU communication paths; shared-storage setup, diagnosis and recovery.

Independent qualification

Workload-based acceptance, reproducible evidence, prioritized findings, controlled changes and technical handover.

Research and IT collaboration

Work with researchers, administrators, procurement and integrators around an agreed scope and operating responsibility.

## Experience across a career More than 100,000 compute nodes and 100,000 GPUs deployed and managed, dozens of petabytes of storage, research pipelines across disciplines, HPC teams and user communities. These are career-wide experience figures, not a statement about one project or Wirewalk's current managed estate. ## A practical starting point Qualification conversation → written scope and deliverable → agreed access and test window → evidence, findings and handover. Purchase orders, invoices and other current options are described at **www.wirewalk.com/order/**. **Explore:** www.wirewalk.com/services/ · www.wirewalk.com/samples/ · www.wirewalk.com/lab/