About this role
Architect and evaluate hardware and software enhancements for AI training systems, including analysis of accelerator/GPU architectures, performance/area/power trade-offs, and codesign with GenAI/LLM requirements. Work with customers to determine hardware requirements, define performance models, and perform pre-silicon performance prediction for ML training workloads.
Skills for this role
Machine LearningLLM architecturesGPU architectureAccelerator architectureNVIDIA GPU architecturesAMD GPU architecturesComputer architectureDigital circuitsHardware simulatorsNetwork-on-Chip (NoC)Memory technologiesCompute-in-MemoryGEMMGEMVPerformance modelingPre-silicon performance predictionPower/performance trade-off analysisHardware-software codesignDistributed trainingDeepSpeedFSDPNVIDIA DGXNVL72TensorFlowPyTorchC++PythonCompetitive AnalysisCNNRNN