Description

Own GPU performance modeling, optimization, and silicon characterization for next-generation architectures.

This role is on-site.

Responsibilities

• Maintain and extend cycle-accurate or approximate simulators for Shader Engines, Cache Hierarchies, Memory Subsystems, and Interconnects.
• Execute simulation experiments to evaluate GPU configurations, scaling limits, and trade-offs, providing data-driven recommendations.
• Profile and optimize performance for generative AI, LLM, and graphics workloads, identifying bottlenecks in compute engines and SoC fabrics.
• Lead early silicon validation, capturing telemetry and correlating results with pre-silicon models to root-cause hardware-software mismatches.
• Partner with compiler and runtime teams to implement microarchitectural optimizations that maximize execution unit utilization.

Required Skills

• 5-10 years of industry experience in silicon performance engineering, GPU modeling, or microarchitecture design.
• Strong proficiency in C++ and Python for simulation and analysis tools.
• Deep understanding of GPU architecture, including SIMD/SIMT models, cache hierarchy, and memory technologies.
• Experience with hardware simulation, silicon characterization, and performance analysis tools.
• Familiarity with Linux tracing frameworks, performance counters, and register dumps.
• Understanding of PCIe, high-bandwidth interconnects, and kernel execution graphs.

Preferred Skills

• Experience with AI/LLM workloads, Transformer models, PyTorch, or Triton.
• Knowledge of graphics pipelines, compute shaders, ray tracing, and rasterization.

Key Skills

Education

Any Graduate