Member of Technical Staff, Data Infrastructure Storage

Fireworks • San Mateo, CA

3d • $175,000 - $220,000

About The Position

Fireworks is seeking a technical staff member to build a high-performance data layer for GPU workloads. The team owns how data is stored, cached, and moved across clouds so training and inference workloads can access it without manual data staging or waiting for data transfers. The system spans parallel file systems, object storage, distributed caches, and key-value stores, positioning datasets, checkpoints, model weights, and KV cache where they are needed at scale.

The role focuses on designing, building, and operating storage and caching infrastructure for AI workloads. Relevant areas of depth include storage systems, distributed or tiered caching, Linux I/O internals, and high-performance networking. Candidates should have deep experience in at least one of these areas and be comfortable working across the others.

Requirements

• Bachelor's degree in Computer Science, Engineering, or a related technical field, or equivalent practical experience.
• At least 5 years of experience designing, building, and operating storage, caching, or data infrastructure in production at scale.
• Strong software development skills in C/C++, Rust, or Go, including experience writing performance-critical data-path code; Python or Go for automation.
• Deep understanding of distributed storage fundamentals, including consistency models, replication and consensus, metadata scaling, caching, and failure recovery.
• Hands-on experience with Kubernetes, including CSI, to mount high-performance volumes into ephemeral GPU pods.
• Experience deploying and operating storage clusters such as WEKA, VAST, Lustre, Ceph, or MinIO across clouds using infrastructure-as-code tools such as Terraform or Pulumi.
• Proficiency with performance benchmarking and profiling tools such as fio, iperf3, NCCL tests, perf, and eBPF.
• Strong depth in at least one of storage systems, distributed or tiered caching, Linux internals, or high-performance networking, with comfort working across the other areas.

Nice To Haves

• Experience as a core contributor or owner of a storage, caching, or key-value system, in-house or open source, such as Ceph, Alluxio, JuiceFS, RocksDB, FoundationDB, TiKV, Redis/Dragonfly, SeaweedFS, or MinIO.
• Experience with high-performance computing or large-scale AI clusters, such as Slurm or Kubernetes GPU pools, and familiarity with AI I/O patterns including checkpoint bursts, random dataset reads, and model-weight loading for inference cold starts.
• Familiarity with Linux internals, including kernel I/O paths, POSIX semantics, page cache behavior, io_uring/SPDK, and OS-level tuning such as sysctl, I/O schedulers, and NUMA.
• Familiarity with high-performance networking, including RDMA, RoCE v2, InfiniBand, and GPUDirect Storage.
• Experience moving, replicating, or caching petabyte-scale datasets across clouds, including AWS, GCP, and specialized neoclouds, with attention to egress cost and consistency.
• Preference for software-defined storage over proprietary hardware appliances.

Responsibilities

• Build a unified, high-performance data layer spanning parallel file systems, object storage, distributed caches, and key-value stores across multiple clouds.
• Enable training and inference workloads to access datasets, checkpoints, model weights, and KV cache without manual data staging or delays.
• Design, build, and operate storage, caching, or data infrastructure in production at scale.
• Deploy and operate storage clusters across clouds, including mounting high-performance volumes into ephemeral GPU pods.
• Benchmark and profile performance to identify and resolve bottlenecks across the kernel, network, cache, and storage backend.
• Automate infrastructure deployment and operations using infrastructure-as-code tools.

What This Job Offers

• Job Type: Full-time
• Career Level: Senior