GPU Cluster Sysadmin

Stanford Research Computing (https://) is a collaboration between University IT and the Vice Provost and Dean of Research. They operate HPC environments for researchers, provide one-time consultations on projects (software and pipelines, data management, physical building design and fit-out), and provide contract support for individual Labs, Departments, and Schools. They have four open positions: three hybrid and one (marked) onsite.

Lead the storage team and set direction for large storage environments Oak (file storage used by multiple clusters), Fir (fast scratch for Sherlock), and Elm (object storage on top of tape). Knowledge of Lustre, Infiniband, and PB-scale storage is important.

Maintain and expand Oak, a 20+ Pebibyte Lustre storage environment used by the largest HPC clusters. Depending on experience, may also have responsibility for Elm (object storage on top of tape). Knowledge of Lustre, Infiniband, and PB-scale storage is important.

With Marlowe—an 1SU NVIDIA DGX H100 SuperPOD with DDN Intelliflash and DDN NFS storage—work on latest AI/ML/Deep Learning/LLM software and frameworks, keep the environment up-to-date, and work with NVIDIA/DDN when trouble occurs.

System administrator to help run hardware/infrastructure portions of Sherlock, their largest HPC cluster. Sherlock includes Intel and AMD  servers, three Infiniband fabrics, and an Ethernet backbone. Responsible for maintaining, troubleshooting, and improving the hardware.

• Relocation incentive is provided if you do not live in the Bay Area
• free transit passes depending on where you live
• parking costs apply if you drive on on-site days
• some on-call around the holidays

• 403(b) match
• good healthcare
• 30+ days off per year (holidays + vacation)

Benefits are documented publicly at https://.

Questions can be asked by replying here or emailing the person (info in profile).