Senior Staff Engineer - AI Workloads & Storage - San Jose, California, United States

Originally posted 27 August 2026 by the employer.

Focus on NAND/SSD storage systems designed for large-scale AI.

About the role

This role involves developing the host interface firmware layer at the intersection of system software and flash management firmware, driving Samsung's V-NAND technology for performance-oriented, demanding, enterprise-class applications. This position is focused on making SSDs first-class citizens in the AI data path, from the NAND media and flash-translation layer up through NVMe and networked storage.

What you'll do

• Own AI workload characterization, profiling LLM inference, RAG, and training workloads to quantify I/O, bandwidth, latency, and capacity demands.
• Identify optimal data placement, analyzing workload access patterns and mapping insights onto SSD data-placement technologies such as NVMe Flexible Data Placement (FDP) and streams.
• Collaborate with customers to identify differentiating SSD capabilities for AI workloads and develop proof-of-concept implementations.
• Lead deep-dive performance analysis spanning inference runtime, the Linux storage and networking stack, and underlying hardware.
• Build and evaluate transactional and system-level models of proposed architectures.
• Engage with standards and the open ecosystem, including SNIA (Storage.AI), MLCommons/MLPerf, and the open inference stack.

What you'll need

• 10–15+ years of experience in systems, storage, or ML-systems software, with a track record of architecting systems.
• Demonstrated technical leadership and cross-team influence, including setting direction and mentoring senior engineers.
• Working knowledge of modern AI inference, especially transformer architectures.
• Deep systems-level understanding of the Linux storage stack (block layer, I/O scheduling, NVMe) and NAND/SSD internals.
• Hands-on performance analysis skills using tools like perf, ftrace, eBPF, blktrace, and fio.
• Fluency in Python plus a systems language (C/C++, Rust, or Go).

Nice to have

• Hands-on experience with the modern inference stack: vLLM, SGLang, LMCache, NVIDIA Dynamo, TensorRT-LLM, or Triton.
• Familiarity with GPU-adjacent data movement and memory frameworks: NIXL, DOCA / DOCA MemOps, GPUDirect Storage, RDMA, NVMe-oF, and BlueField / DPU offload.
• Understanding of GPU and TPU architecture.
• Experience with user-mode storage access frameworks: SPDK, uNVMe, libvfn, or similar.
• SSD firmware experience, including flash-translation layer, wear-leveling, garbage-collection algorithms, and data-placement features like FDP / streams / ZNS.
• AI-workload characterization and benchmarking experience, and familiarity with SNIA Storage.AI and MLCommons / MLPerf.
• Transactional / discrete-event or system-level modeling experience in frameworks such as SystemC, SimPy, or similar.
• Experience with SSD architecture and interfaces: NVMe (including ZNS, Flexible Data Placement / FDP), open-channel SSDs, computational storage; and with PCIe Gen5, CXL, and large-scale GPU-cluster storage (VAST, WEKA, Lustre, Ceph).

Skills: AI workloads, storage systems, NAND/SSD, Linux storage stack, NVMe