Principal Agentic AI Performance Optimization Engineer

The focus of this role is agentic systems that enable end-to-end optimization of AI workload performance on AMD hardware. You will join a team developing tools that support the broad adoption of agentic optimization loops to safely and reliably optimize AI workloads. This opportunity combines platform and tooling engineering with hands-on work in inference serving and agentic systems.

Develop and enhance tooling that improves the developer and operator experience across the performance optimization stack. Collaborate with researchers and engineers to transform optimization capabilities into reliable, well-tested, and well-documented solutions.

Create safety and validation frameworks, in partnership with stakeholder teams, for agentic systems that propose and apply source code and configuration changes.

Design flexible abstractions that support a variety of workloads, hardware platforms, and cluster environments.

Package validated optimizations into versioned, deployable artifacts with clear provenance and traceability.

Drive adoption and distribution of shared tooling components across teams and repositories, integrating with public framework interfaces and contributing improvements where appropriate.

We recognize that candidates may bring valuable experience through different paths. If your background aligns with many of the areas below, we encourage you to apply.

Experience building developer tooling, command line tools, or automation platforms in Python, ideally within infrastructure, MLOps, or performance engineering environments.

Experience deploying LLMs or other AI models in production using frameworks such as vLLM, SGLang, or similar technologies.

Familiarity with agentic systems and AI-driven workflows.

Demonstrated success designing scalable software solutions.

Experience with containerization technologies such as Docker and cluster scheduling platforms.

Curiosity and enthusiasm for learning in a rapidly evolving technology landscape.

A collaborative approach, practical problem-solving skills, and the initiative to drive solutions forward.

Degree in Computer Science, Computer Engineering, Electrical Engineering, or a related field, or equivalent practical experience.

Experience designing systems with observability and auditability in mind.

Experience building CI/CD solutions, particularly for validating and qualifying AI workload optimizations.

Familiarity with validation, gating, or decision-making frameworks related to AI workload performance optimization.

Experience with Kubernetes and Helm.

Knowledge of GPU architecture, kernel development, and debugging using technologies such as C, C++, HIP, or CUDA.

Experience using AI-assisted software development tools and workflows.