Opens job-boards.greenhouse.io in a new tab

Overview

  • To provide the best candidate experience amidst our high application volumes, each candidate is limited to 10 applications across all open jobs within a 6-month period. 
  • Advancing the World’s Technology Together
  • Our technology solutions power the tools you use every day--including smartphones, electric vehicles, hyperscale data centers, IoT devices, and so much more.
  • Here, you’ll have an opportunity to be part of a global leader whose innovative designs are pushing the boundaries of what’s possible and powering the future. 
  • We believe innovation and growth are driven by an inclusive culture and a diverse workforce.
  • We’re dedicated to empowering people to be their true selves.
  • Together, we’re building a better tomorrow for our employees, customers, partners, and communities.
  • What You’ll Learn
  • Project Overview: Build AgentSysBench, a benchmark suite and profiling toolchain that measures how AI agent frameworks (Claude Code, Aider, LangGraph) use server resources: host CPU, DRAM, GPU HBM and CXL memory.
  • The results will guide memory system design for future AI data centers.
  • Skills You’ll Learn:
  • Profile system and microarchitecture performance with perf, eBPF, hardware PMUs and NVML/DCGM
  • Characterize memory hierarchies (HBM, DDR5, CXL) and build performance models
  • Benchmark LLM inference serving (vLLM/SGLang) on NVIDIA GPU servers
  • Our team researches the next-generation memory solutions for AI workloads, including HBM, DRAM, and CXL memory expansion.
  • This summer, you'll work on servers with Samsung CXL memory expanders and NVIDIA GPUs.
  • You'll see how agentic AI workloads stress memory systems and how those findings feed into future memory product architecture design.
  • Location: Daily onsite presence at our San Jose headquarters in alignment with our Flexible Work policy
  • Reports to: Sr.
  • Build runtime adapters and a trace record/replay harness for AI agent frameworks
  • Develop micro-kernel and end-to-end agent workload benchmarks that stress CPU, memory, storage and accelerator subsystems
  • Profile agent workloads with perf, eBPF, PMUs and NVML/DCGM to break down per-turn latency and memory bandwidth usage
  • Measure host DDR5, GPU HBM and CXL.mem bandwidth and latency across Intel Xeon and/or AMD EPYC servers
  • Evaluate multi-agent KV-cache sharing and CXL memory expansion on Samsung CXL expanders
  • Build an initial memory tiering cost model (HBM/DDR5/CXL) for long-running MoE agent workloads
  • Present findings in weekly team meetings, then deliver a final report and presentation

What You'll Do

  • Pursuing PhD in Computer Science, Computer Engineering, Electrical Engineering, or a related field
  • Strong programming skills in Python and C/C++ in a Linux environment
  • Solid understanding of computer architecture (CPU caches, the memory hierarchy, NUMA)
  • Experience with performance profiling tools (perf, eBPF, VTune, or Nsight)
  • Preferred: hands-on experience with LLM serving frameworks (vLLM, SGLang) or GPU programming
  • Preferred: familiarity with HBM, CXL, or heterogeneous memory systems
  • Preferred: experience with AI agent frameworks or benchmark development
  • You’re inclusive, adapting your style to the situation and diverse global norms of our people.

Sourced directly from Samsung Semiconductor’s career page

Your application goes straight to Samsung Semiconductor.

Samsung Semiconductor logo

Samsung Semiconductor

San Jose, California, United States

Specialisation
Open roles at Samsung Semiconductor
70 positions
Job ID
8202927003

Get matched to roles like this

Upload your resume once. We’ll notify you when matching roles open up.

Join talent pool — free

Similar Other roles