Research Intern - AI System Architecture Modeling and Performance

Microsoft•Hillsboro, OR
•$6,810 - $14,610•Onsite

About The Position

Microsoft Research Internships offer a dynamic environment for research careers, connecting interns with world-class research labs and researchers working across scientific and technical disciplines. The Azure Hardware and Systems Infrastructure organization defines Microsoft's first-party AI infrastructure architecture and strategy, including SoC designs, interconnect topologies, and memory hierarchies, and contributes to roadmap development from concept through silicon and hyperscale integration. In this 12-week internship, the Research Intern will work at the intersection of hardware and software co-design, evaluating how to optimize AI systems and assessing their real-world impact on Azure hyperscale infrastructure. The role focuses on opportunities to co-optimize CPU, GPU, and networking infrastructure for the Maia accelerator ecosystem. The intern will identify system stress points, propose architectural ideas, and develop methodologies combining workload characterization, modeling, and benchmarking to assess effectiveness. Interns are paired with mentors and collaborate with researchers and fellow interns, present findings, and contribute to the research community.

Requirements

  • Currently enrolled in a Doctorate Degree program in Computer Science or a related STEM field.
  • Be physically located at the manager’s Microsoft worksite for the duration of the internship.
  • Submit a minimum of two reference letters, a cover letter, and relevant work or research samples.
  • Have solid verbal and written communication skills.

Nice To Haves

  • At least 1 year of experience or equivalent coursework related to performance analysis tools and methodologies, optimization, and modeling.
  • Proficiency with inference-serving frameworks such as SGLang, vLLM, and Dynamo, as well as PyTorch and AI accelerator programming models or compilers such as CUDA and Triton.
  • Deep understanding of GPU and AI architectures, including memory hierarchies, compute-communication interplay, kernel scheduling, and interconnect properties.
  • Familiarity with CPU/server architectures, including PCIe topologies and accelerator, NIC, and peripheral demand; solid understanding of CPU involvement in dispatching, scheduling, and orchestration of input data pipelines to AI accelerators.
  • Hands-on experience with benchmarking, profiling, identifying performance bottlenecks, and performance analysis and optimization, including trace generation, event monitoring, and instrumentation.
  • Familiarity with roofline performance modeling, detailed performance simulations, and speed-versus-accuracy tradeoffs in performance modeling methodologies.
  • Ability to apply appropriate performance analysis methodologies, including devising new or combined approaches to evaluate complex system architecture what-if scenarios.

Responsibilities

  • Evaluate opportunities to co-optimize CPU, GPU, and networking infrastructure for the Maia accelerator ecosystem.
  • Identify system stress points and propose novel architectural ideas.
  • Create methodologies using workload characterization, modeling, and benchmarking to evaluate architectural ideas.
  • Collaborate with mentors, other Research Interns, and researchers.
  • Present findings and contribute to the research community.
© 2026 Teal Labs, Inc
Privacy PolicyTerms of Service