AI/ML Memory Performance Intern: Trace & Bottleneck Analysis

Micron Technology, Inc.; San Jose (CA); Presencial; USD 69.000 - 77.000; Jornada completa

Descripción de la vacante

Micron Technology, Inc. is seeking an AI/ML Workload Analysis Intern to join our performance and architecture teams in the US.

You will analyze AI workloads, develop memory trace methods, and uncover bottlenecks across CPUs/GPUs/NPUs and memory systems, contributing to next-generation memory technologies. You will work with system architects, performance engineers, and researchers to collect traces, build models, and validate simulations, gaining hands-on experience in memory-centric AI

Formación

• Pursuing MS or PhD in related field with focus on computer architecture or ML.
• Experience in programming languages and performance analysis.
• Strong analytical and communication skills.

Responsabilidades

• Analyze AI/ML workloads including LLMs, recommender models, and vision models.
• Collect, process, and analyze memory traces from AI workloads.
• Characterize memory access patterns, bandwidth, and latency sensitivity.
• Develop workload traces, performance models, and simulation environments.
• Identify system bottlenecks related to memory and interconnects.
• Validate models with benchmarks and hardware measurements.
• Present findings through technical reports and reviews.

Conocimientos

Python

C/C++

Analytical thinking

Communication skills

Educación

Pursuing MS or PhD in Computer Engineering, Computer Science, Electrical Engineering, Data Science, or a related field

Herramientas

Nsight

VTune

perf

rocprof

Gem5

Accel-Sim

Linux

Descripción del empleo

Micron Technology, Inc. is seeking an AI/ML Workload Analysis Intern to join our performance and architecture teams in the US.

You will analyze AI workloads, develop memory trace methods, and uncover bottlenecks across CPUs/GPUs/NPUs and memory systems, contributing to next-generation memory technologies. You will work with system architects, performance engineers, and researchers to collect traces, build models, and validate simulations, gaining hands-on experience in memory-centric AI