Ryan Glusic / Engineer & builder

Curiosity,put to work.

IT DevOps Engineer at NVIDIA, running Slurm workload orchestration for the HPC fleet and building AI-powered operations tools.

Explore the work View resume

Infrastructure / AI agents / Exploration

FIG. 001 / CONTINUOUS FORMX · Y · Z
A small study in perspective
Working acrossHPC & GPU systemsLinux & orchestrationScientific computing
/01Selected work

Agents. Systems. Experiments.

Current work in AI agents and the infrastructure underneath them.

Current work · Built for NVHPCIn development

An interactive DevOps assistant built for NVHPC to automate routine operational workflows. I actively use it through Slack to request system reboots and inspect Slurm and job pools.

The need: offload routine operational work while maintaining round-the-clock awareness. The design aims to discover relevant context as it becomes familiar with the environment, answer node-health questions, and generate dynamic, contextual reports instead of static ones.

In development: automated failure diagnosis and diagnostic job reruns.

Core capabilities
  • 01System reboot requests through Slack
  • 02Conversational Slurm & job-pool inspection
  • 03Slurm monitoring & issue alerts
  • 04Persistent memory & consolidation
  • 05Administrator approval & denial gates
Internal work / Designed for team use
Information in motion
Abstract information-flow artwork

Current contributions / NVIDIA

The infrastructure behind the agents.

Fleet observability

Created and maintain the fleetwide Prometheus alerting system and Grafana observability stack for GPU/HPC infrastructure — heterogeneous clusters of 150+ nodes, including DCGM exporter coverage.

Workload orchestration

Own the Slurm configuration stack for the entire fleet — one of three engineers holding root access. The fleet is heterogeneous: x86 and Arm64, NVIDIA Grace to AWS Graviton, GPU generations from Tesla onward. Co-manage the Ansible automation around it.

Testing tools

Build and improve the NVHPC testing tools the team's validation runs on.

System recovery

Respond in person to lab system failures — kernel, driver, and GPU troubleshooting through to full recovery.

/02The throughline

A physics foundation.
A systems mindset.

My work brings together GPU infrastructure, Linux, workload orchestration, and AI agent design. My background in physics and mathematics connects that work to scientific computing and software.

Alongside engineering, I continue to explore AI research.

B.S. in Physics and Mathematics
University of Wisconsin–Parkside · 2021

01

AI agents

Workflows with oversight.

Interactive DevOps assistants that automate routine workflows through infrastructure tools, persistent memory, and administrator approval controls.

Workflow automationMemory lifecycleHuman approval
02

Infrastructure

The systems underneath.

GPU/HPC fleet operations at root level: Slurm orchestration, Prometheus and Grafana observability, Ansible automation, and hands-on recovery — plus the team's AWS and Azure work.

SlurmLinuxPrometheusGrafanaAnsibleAWSAzure
03

Software

Ideas made executable.

NVHPC testing tools, scheduler wrappers, and scientific workflows — Python that runs against real systems.

PythonToolingScientific computing
/03From the archive

Where the questions started.

Browse the research archive
/04Make contact

Let's compare notes.

Infrastructure, software, scientific computing.
Start a conversation.

View my resume