Downloaded from martin.roch-neirey.com
Martin Roch-Neirey.
Network & Systems Engineer · Montréal, Canada
Profile
Network and systems engineer and student at Polytechnique Montréal, specialized in running GPU clusters on Slurm and Kubernetes at scale (NVIDIA GPUs, CUDA, NVIDIA GPU Operator, vGPU), backed by HPC, NetDevOps and cloud, with security in mind. Co-founder of Glow Up.
Skills
- Slurm
- Kubernetes
- GPU Management
- CUDA
- HPC
- vSphere / vGPU
- Docker
- Terraform
- Ansible
- Cloud & virtualization
- NetDevOps
- Linux
- Python
- Go
Selected projects
- Slurm HPC cluster
Designed and operated the first Slurm cluster of my former school, scheduling NVIDIA GPUs (GH200, A40, A6000) for many data scientists. Full CUDA and NVIDIA Container Toolkit setup, bare-metal and vSphere vGPUs with cloud bursting, three storage spaces.
- Kubernetes GPU cluster rebuild
Full rebuild of a GPU cluster (3 control-plane + 15 workers) exposing NVIDIA GPUs (GH200, A40, A6000) via the NVIDIA GPU Operator, CUDA and the Container Toolkit. Bare-metal and vSphere vGPUs, cloud bursting, Ceph CSI, Rancher. Inference platform.
- Carbon-aware scheduling (FLUIDOS)
Carbon-aware workload shifting across geo-distributed Kubernetes clusters. Paper under review at IEEE Transactions on Sustainable Computing.
- Benchmarking cluster compute capacity
Terraform, Ansible and High Performance Linpack to measure and compare vSphere, OpenStack and Kubernetes.
- Glow Up: SaaS & mobile app
Co-founder. A free SaaS for beauty & wellness professionals (operational and strategic steering), paired with a client mobile app.
Education
- Master of Engineering, Computer Science & Communication Systems Polytechnique Montréal · Montréal, Canada2025
- Bachelor of Science, Computer Science & Communication Systems Haute école d'ingénierie et d'architecture de Fribourg · Fribourg, Switzerland2024
- Technological University Diploma, Computer Networks & Telecommunications Université Savoie Mont-Blanc · Annecy, France2021