$152,000 - $287,500 USD yearly
Build data platform infrastructure and storage tiering on Kubernetes for Managed AI Research Superclusters (MARS).
Originally posted November 9, 2026 by the employer.
About the role
The MARS AI Data team owns how AI research uses data across multi-cloud GPU fleets, covering datasets, checkpoints, and scratch storage across file systems, object storage, and local NVMe. You will help design the MARS data platform on Kubernetes from the ground up, working directly with researchers training foundational models.
What you'll do
- Design and build the AI data layer for a multi-cloud GPU fleet, managing data placement, caching, tiering, and movement across storage backends.
- Build high-throughput data paths for training and reinforcement learning, including streaming loaders from object storage, distributed caches, and checkpoint save and restore at thousands of GPUs.
- Construct Kubernetes-native APIs, operators, and storage integrations to mount, cache, and account for job data across Slurm and Kubernetes platforms.
- Manage quotas, capacity, retention, and lifecycle enforcement across research teams, including usage attribution and capacity forecasting.
- Build I/O telemetry and access analytics to monitor metadata pressure, I/O stalls, and hot/cold data access patterns.
- Own production reliability, including on-call duties, incident analysis, and systemic mitigations for issues like full file systems and lost checkpoints.
What you'll need
- BS or MS in Computer Science, Computer Engineering, Electrical Engineering, a related field, or equivalent experience.
- 5+ years of experience building production systems or data-infrastructure software.
- Deep experience with parallel or distributed file systems (Lustre, WEKA, VAST, GPFS, Ceph, or similar), object storage at scale (S3 or compatible), or large-scale data pipelines and data lakes.
- Strong understanding of I/O performance, including throughput, latency, metadata cost, caching, and access patterns.
- Strong programming skills in Go, Rust, C, or C++, alongside Python for tooling and analysis.
- Demonstrated track record of owning a production service and improving its reliability and performance.
Skills: distributed file systems, Kubernetes, checkpointing, object storage, NVMe
More semiconductor jobs: NVIDIA jobs
Tracked in our index since 10 Nov 2026. This is our observation date, not the employer posting date. Specialty: Infrastructure/Platform Software (Storage Systems). 39 active roles in this specialty at 13 employers in our index, as of 19 Nov 2026.
How we counted: postings read from employer career sites. How we count