Staff Engineer - AI Workload Benchmarking

Xoriant; San Jose (CA); On-site; USD 140,000 - 210,000; Full time

Job summary

Xoriant is seeking a highly skilled Systems Performance Engineer to drive benchmarking for memory and storage leadership aligned with roadmap needs. The role emphasizes performance-driven system design and validation for next-generation AI infrastructure, including AI workloads and hardware accelerators.

Ideal candidates will have deep Linux performance expertise, NVMe storage experience, and the ability to produce reproducible benchmark artifacts and technical reports for senior engineers.

Qualifications

• Experience in systems performance engineering and storage benchmarking.
• Expertise with NVMe SSDs and storage stack performance analysis.
• Hands-on with AI/ML workloads and LLM frameworks.
• Proficiency with Linux performance tracing tools (blktrace, perf, bpftrace, etc.).
• Ability to deliver reproducible benchmark artifacts and reports to senior engineers.

Responsibilities

• Align benchmarking insights to leadership in memory and storage to inform roadmaps.
• Enable AI infrastructure solutions through performance-driven system design and validation.

Skills

Systems perf eng

Storage performance

AI infra benchmarking

Python

C/C++ basics

Education

Bachelor's/Master's degree in CS/CE/EE

Tools

blktrace

perf

eBPF/bpftrace

ftrace

iostat

fio

Job description

• Align benchmarking insights to Samsung's leadership in memory (HBM, DRAM, CXL) and storage (SSD/NAND) to inform product roadmaps.
• Enable next-generation AI infrastructure solutions through performance-driven system design, validation, and standards engagement.

Requirements

• Bachelor's/ Master's degree in Computer Science, Computer Engineering, Electrical Engineering, or a related field; advanced degree preferred.
• Professional experience in systems performance engineering, storage performance, or AI infrastructure benchmarking.
• Demonstrated expertise with NVMe SSDs and storage stack performance analysis (block layer, page cache, file systems, asynchronous I/O).
• Hands-on experience with AI/ML workloads — LLM training and inference frameworks (PyTorch, vLLM, TensorRT-LLM, or equivalent), embedding pipelines, or vector databases (FAISS, Milvus, DiskANN, HNSW).
• Strong proficiency with Linux performance and tracing tools: blktrace, perf, eBPF/bpftrace, ftrace, BCC, iostat, fio.
• Working knowledge of GPU systems and accelerator I/O paths
• Experience designing and executing benchmarks against industry standards (MLPerf Storage, or equivalent)
• Proficiency in Python for benchmarking automation, data analysis, and visualization; comfort with C/C++ for systems-level work.
• Proven ability to deliver structured technical reports, characterization studies, and reproducible benchmark artifacts to a senior engineering audience.