Senior Staff Engineer - AI Workloads & Storage
Samsung Semiconductor1 day ago
San Jose, CA, USAStaff+
Base Salary
$189k - $301k/yr
Responsibilities
- Characterize production and emerging LLM inference, RAG, and training workloads to quantify I/O, bandwidth, latency, and capacity requirements.
- Determine optimal data placement on flash and apply NVMe Flexible Data Placement and streams to improve write amplification, endurance, latency, and QoS.
- Collaborate with customers to identify differentiated SSD capabilities and develop proof-of-concept data-path, tiering, and data-placement implementations.
- Analyze performance across inference runtimes, Linux storage and networking, and underlying hardware, tuning latency, throughput, cost, and GPU utilization.
- Build and validate transactional, discrete-event, and system-level models of proposed architectures.
- Engage with SNIA Storage.AI, MLCommons/MLPerf, and the open inference ecosystem.
- Set technical direction, make build-versus-buy and architecture decisions, establish benchmarking practices, and mentor engineers.
- Partner with product, hardware, research, vendors, and external partners to move architectures from concept to deployment.
Requirements
- Bachelor's degree with 15+ years of relevant industry experience, Master's degree with 13+ years, or PhD with 10+ years of relevant industry experience.
- Typically 10–15+ years of experience in systems, storage, or ML-systems software with a record of materially improving performance, reliability, or cost.
- Demonstrated technical leadership, cross-team influence, decision-making across organizational boundaries, and mentoring experience.
- Working knowledge of modern AI inference and transformer concepts including attention, KV cache, batching, and memory/compute trade-offs.
- Deep understanding of the Linux storage stack, NVMe, NAND/SSD internals, flash-translation layers, garbage collection, endurance, write amplification, and latency behavior.
- Hands-on performance-analysis experience with tools such as perf, ftrace, eBPF, blktrace, and fio.
- Fluency in Python and a systems language such as C, C++, Rust, or Go.
- Preferred qualifications include an MS or PhD in Computer Science, Electrical/Computer Engineering, or a related field, or equivalent practical experience.
- Preferred experience includes vLLM, SGLang, LMCache, NVIDIA Dynamo, TensorRT-LLM, Triton, NIXL, DOCA, DOCA MemOps, GPUDirect Storage, RDMA, NVMe-oF, BlueField, DPU offload, SPDK, uNVMe, libvfn, SSD firmware, FDP, streams, ZNS, SNIA Storage.AI, MLCommons/MLPerf, SystemC, SimPy, NVMe, open-channel SSDs, computational storage, PCIe Gen5, CXL, VAST, WEKA, Lustre, and Ceph.
Benefits
- Base benefits include medical, dental, vision, and 401(k) coverage.
- Charitable giving match and community involvement opportunities are offered.
- The role includes 4+ weeks of paid time off annually, plus holidays and sick leave.
- Family support includes fertility-care or adoption stipends, medical travel support, and virtual veterinary care.
- Emotional wellness benefits include on-demand apps and free confidential therapy sessions.
- Onsite café, gym, and virtual fitness classes are available.
- The company offers a flexible work environment.