Performance Architect

Responsibilities

• Build SystemC performance models for AI Storage Solutions covering end-to-end from GPU/TPU/NPU/xPU to memory hierarchy
• Improve AI/ML ASIC Architecture performance through hardware & software co-optimization and post-silicon analysis
• Perform workload analysis and characterization of ASIC and competitive datacenter solutions
• Model and optimize performance for multi-trillion parameter LLM training and inference
• Architect memory-efficient training systems using techniques like quantization and speculative decoding
• Collaborate with internal and external stakeholders to iterate on architectural designs

Requirements

• Bachelor's with 7+ years, Master's with 5+ years, or PhD with 3+ years in Performance Modeling/Simulation/Analysis
• 5+ years of experience with SystemC modeling
• Experience with simulation using SystemC and TLM, behavioral modeling, and performance analysis
• Strong understanding of computer/graphics architecture, ML, and LLM

Preferred Qualifications

• Experience with storage systems, protocols, and NAND flash
• Deep experience optimizing large-scale ML systems and GPU architectures
• Expert knowledge of transformer architectures and model parallelism
• Proficiency in PyTorch, CUDA, TensorRT, OpenAI Triton, and ONNX
• Experience with distributed systems like Ray or Megatron-LM
• Familiarity with NVMe, PCIe, UCIe, CXL, or NVLink protocols

About the Company

Sandisk understands how people and businesses consume data and relentlessly innovates to deliver solutions that enable today’s needs and tomorrow’s next big ideas through groundbreaking innovations in Flash and advanced memory technologies.