Dc-gpu Performance Modeling Engineer

AMD AMD · Semiconductors · Bangalore, India · Engineering

This role focuses on architecting, analyzing, and optimizing high-performance GPU-centric SoCs for Machine Learning workloads. The engineer will develop performance models and methodologies, propose solutions for performance and power optimization, and collaborate with various teams on hardware/software co-design for next-generation data center systems. Experience with ML models, distributed training, and inference is preferred.

What you'd actually do

  1. Define, build and maintain performance models for performance projections, analysis and architecture exploration.
  2. Develop and execute system-level modelling strategies for ML and GPU hardware and software co-design.
  3. Drive performance trade-off studies for new architectural features, algorithms, and system configurations, providing data-driven recommendations.
  4. Collaborate with architecture, design and software teams to integrate models, define workloads and analyse simulation results.
  5. Innovate and advance modelling methodologies, tools and infrastructure to improve accuracy, speed, and architectural insight.

Skills

Required

  • Python
  • performance analysis
  • hardware/software co-design
  • system-level modelling
  • ML ops
  • parallelization strategies
  • computer architecture

Nice to have

  • GPUs
  • SoCs
  • ML accelerators
  • workload characterization
  • ML models and software stacks
  • AI model distributed training and inference

What the JD emphasized

  • performance models
  • performance analysis
  • ML models
  • AI model distributed training and inference

Other signals

  • GPU-centric SoCs for Machine Learning workload
  • performance models and methodologies
  • optimize power for next-generation data centre systems
  • ML and GPU hardware and software co-design
  • performance trade-off studies
  • AI model distributed training and inference