Software Development Engineer 5

AMD is building agentic software to improve how teams develop, deploy, and operate machine learning workloads on AMD GPUs. This senior contract role designs and delivers production-grade intelligent agents across AMD’s software stack, automating engineering work in ML operations, distributed job troubleshooting, and performance analysis.

Responsibilities:

• Architect and implement Python agent frameworks, including:
• Tool adapters
• Orchestration loops
• Structured outputs
• Session handling
• CLI or service interfaces suitable for production use
• Build LLM-powered workflows using LiteLLM, DSPy, or equivalent
• Implement bounded prompts, allowlisted tools, budget caps, structured JSON outputs, and evidence-backed reporting
• Integrate agents with the AMD GPU and ML software ecosystem, including:
• ROCm tooling
• PyTorch
• Distributed training and inference stacks
• Profilers
• Cluster metrics
• Log pipelines
• Debug utilities
• Solve ML operations problems on training and inference clusters, including:
• Multi-node failure analysis
• Performance regression investigation
• Configuration and launch issues
• Operational automation
• Own agent quality and evaluation infrastructure
• Develop fixture-driven tests and regression suites
• Implement JSON Schema validation and accuracy metrics
• Integrate testing and evaluation into CI before rollout
• Deliver deployment-ready artifacts, including:
• Docker packaging
• Kubernetes packaging
• Runbooks
• Technical documentation
• Apply security-conscious defaults for customer-controlled environments
• Partner with framework, performance, and field teams
• Incorporate code review feedback and iterate based on production and pilot learnings
• Own significant technical components independently and deliver production-quality solutions with minimal supervision

Requirements:

• Bachelor's degree or higher in:
  • Computer Science
  • Computer Engineering
  • Electrical Engineering
  • Related field
• Equivalent practical experience accepted
• 10+ years of professional software engineering experience
• 5+ years of experience developing production Python systems at scale
• Deep experience with distributed ML training or inference on GPU clusters
• Multi-node jobs
• Collective communication failures
• Log and metric correlation across ranks and nodes
• Strong hands-on PyTorch experience
• Production familiarity with large-scale ML stacks such as:
  • Megatron-LM
  • DeepSpeed
  • TorchTitan
  • vLLM
  • Equivalent technologies
• Proven ML debugging and performance engineering experience across multiple failure modes, including:
  • Throughput regression
  • Memory errors
  • Numerical instability
  • Misconfiguration
  • Profiler or trace analysis
• Demonstrated experience delivering LLM agent or orchestration systems in production or near-production environments
• Tool routing
• Structured outputs
• Reliability
• Testability
• Track record of delivering complex software on schedule
• Clean, maintainable code
• Automated testing
• Code review
• Technical communication
• Ability to work independently, prioritize ambiguous requirements, and align weekly with a technical lead
• Open-source ML development, including:
  • Upstream framework repositories
  • Community CI patterns
  • OSS tooling integration
• ML performance optimization, including:
  • Parallelism
  • Throughput/MFU tuning
  • Roofline analysis
  • Profiler-driven investigation
• ROCm and GPU cluster operations, including:
  • Metrics
  • Profilers
  • Health monitoring
  • Slurm
  • Kubernetes job environments
• Security-aware agent deployment, including:
  • Secret handling
  • Egress control
  • Customer VPC constraints
  • On-premises deployment constraints