Sw Solution Engineer-ai

AMD AMD · Semiconductors · Beijing, China · Engineering

Software Solution Engineer at AMD focused on optimizing complex application software, particularly AI workloads, on AMD's server platform. Responsibilities include diagnosing and resolving performance issues, optimizing Linux system configurations, analyzing system behavior, and leading AI workload optimization for inference and utilization. The role involves performance testing, benchmarking, and working with AI/ML inference frameworks and quantization techniques.

What you'd actually do

  1. Partner with customers to diagnose and resolve performance issues across applications, operating systems, AI frameworks
  2. Optimize Linux system and kernel configurations to improve application performance while ensuring stability
  3. Analyze system behavior (CPU, memory, resource contention, latency); recommend code and configuration improvements
  4. Use diagnostic tools to identify bottlenecks in computing, memory, storage, and scheduling
  5. Lead AI workload optimization: tune ML frameworks, optimize inference and utilization
  6. Own performance testing and benchmarking: define metrics, build repeatable test environments, validate improvements and document recommendations

Skills

Required

  • Python
  • C++
  • Linux performance analysis tools
  • system debugging
  • Linux OS/kernel experience
  • AI/ML Inference frameworks (vLLM, SGLang, Llama.cpp)
  • profiling tools (e.g., PyTorch Profiler)
  • GPU/CPU architecture
  • quantization (AWQ, GPTQ, GGUF)

Nice to have

  • open-source software development processes and tools
  • debuggers
  • source code control systems (GitHub)
  • profilers
  • compilers

What the JD emphasized

  • performance engineering
  • AI optimization
  • AI/ML Inference frameworks
  • quantization

Other signals

  • optimize complex application software on our server platform
  • delivering performance improvements across server systems, operating systems and AI workloads
  • Lead AI workload optimization: tune ML frameworks, optimize inference and utilization
  • Own performance testing and benchmarking