Principal AI Systems & Performance Lab Architect

SanDisk; Milpitas (CA); On-site; USD 202,000 - 344,000; Full time

Job summary

Sandisk in Milpitas, CA is seeking a Principal Engineer to lead an AI Systems & Performance Lab, directing hands-on work across real-world workloads and diverse hardware. The role focuses on defining the lab's technical vision and delivering actionable insights into system architecture.

You will collaborate with hardware, platform, and product teams to translate workload findings into roadmaps, influence silicon direction, and drive performance improvements across the stack.

Qualifications

• Bachelor’s degree in Computer Engineering, Computer Science, Electrical Engineering or related field; Master’s or PhD strongly preferred.
• 15+ years of industry experience in systems engineering, AI infrastructure, performance engineering, computer architecture, or hardware/software co-design.
• Deep hands-on experience with AI workloads (training and/or inference), including modern deep learning models and frameworks (e.g., PyTorch, TensorFlow, JAX).
• Strong understanding of end-to-end system performance, including compute architectures, memory hierarchy, interconnects, storage, and software stacks.
• Proven ability to identify, analyze, and root-cause performance bottlenecks across heterogeneous systems (CPU, GPU, NPU, accelerators).
• Demonstrated technical leadership with director-level influence, driving strategy and outcomes without large teams.

Responsibilities

• Establish and articulate a clear technical vision and charter for the AI Lab focusing on workload characterization and performance bottleneck analysis.
• Define the research agenda, success metrics, and execution model aligned with organizational objectives and future technology directions.
• Serve as the technical authority for AI systems performance and workload analysis.
• Drive end-to-end analysis of AI training and inference workloads across models, frameworks, runtimes, and full stack.
• Identify performance bottlenecks across compute, memory hierarchy, interconnects, storage, power, and software layers.
• Develop repeatable methodologies for root-cause analysis across heterogeneous platforms (CPU, GPU, NPU, accelerators).
• Architect and oversee benchmarking, profiling, and workload characterization frameworks.
• Evaluate AI performance across hardware configurations and system topologies for rigor and reproducibility.
• Establish best practices for performance measurement, regression tracking, and comparative analysis.

Skills

AI workloads

Bottleneck analysis

System performance

Technical leadership

Executive communication

Education

Bachelor’s degree in related field

Master’s or PhD preferred

Tools

PyTorch

TensorFlow

JAX

GPUs/NPUs

Job description

Sandisk in Milpitas, CA is seeking a Principal Engineer to lead an AI Systems & Performance Lab, directing hands-on work across real-world workloads and diverse hardware. The role focuses on defining the lab's technical vision and delivering actionable insights into system architecture.

You will collaborate with hardware, platform, and product teams to translate workload findings into roadmaps, influence silicon direction, and drive performance improvements across the stack.