Staff Engineer: GPU-Scale Distributed Storage & AI Infra

Fdetimes; San Francisco (CA); Remote; USD 250,000 - 330,000; Full time

Benefits offered by this job

Startup equity

Health insurance

Remote work flexibility

Flexible benefits

Job summary

Together AI seeks an experienced storage engineer to design and operate scalable AI storage infrastructure for multi-petabyte workloads. You will optimize end-to-end data paths, build caching tiers, and integrate advanced storage technologies to support training and inference at scale.

You will architect GPU-focused storage architectures, implement KV-cache offload, and drive multi-tenant isolation while contributing to open-source projects and internal tooling.

Qualifications

• 8+ years in storage engineering at multi-petabyte scale.
• Proven track record building and operating high-performance storage for GPU/HPC.
• Experience scaling for inference: KV-cache offload or similar.
• Strong Go and Python coding for production-grade systems.

Responsibilities

• Architect and implement storage roadmap for Together AI at scale.
• Engineer multi-petabyte AI/ML storage with Vast, Weka, Ceph, and Lustre.
• Develop caching and tiered storage for extreme IOPS and throughput.
• Build KV-cache offload across GPU memory, NVMe, and distributed cache.
• Tune isolation for multi-tenant storage in production.
• Code Kubernetes storage operators and quota enforcement.

Skills

Go

Python

Kubernetes

Distributed storage

GPU/HPC

Performance optimization

Education

BS/MS in CS/Engineering

Tools

Ceph

WekaFS

Lustre

Vast

GPFS

S3/MinIO

Job description

Together AI seeks an experienced storage engineer to design and operate scalable AI storage infrastructure for multi-petabyte workloads. You will optimize end-to-end data paths, build caching tiers, and integrate advanced storage technologies to support training and inference at scale.

You will architect GPU-focused storage architectures, implement KV-cache offload, and drive multi-tenant isolation while contributing to open-source projects and internal tooling.