Principal Engineer - AI Framework Integration

Advanced Micro Devices, Inc

25d•Hybrid

About The Position

At AMD, our mission is to build great products that accelerate next-generation computing experiences—from AI and data centers, to PCs, gaming and embedded systems. Grounded in a culture of innovation and collaboration, we believe real progress comes from bold ideas, human ingenuity and a shared passion to create something extraordinary. When you join AMD, you’ll discover the real differentiator is our culture. We push the limits of innovation to solve the world’s most important challenges—striving for execution excellence, while being direct, humble, collaborative, and inclusive of diverse perspectives. Join us as we shape the future of AI and beyond. Together, we advance your career. THE ROLE As Principal Software Engineer, you will act as the technical authority responsible for end‑to‑end integration of the ROCm software libraries stack and AI frameworks, including PyTorch, TensorFlow, Jax, ONNX, and others. You will operate at the intersection of GPU libraries, compilers, frameworks, and performance tooling, ensuring that our core ROCm library components—such as HIP libraries, Composable Kernel, MIOpen, and hipDNN—are delivered in a form that is correct, performant, and consumable by framework teams and external users. This is a cross‑organizational leadership role. You will not own all code, but you will own integration outcomes, defining what “done” means for framework enablement, performance parity, influencing, adjusting and impacting the roadmap of our core library teams. THE PERSON You are a deeply experienced software engineer who understands that integration is a first‑class engineering problem. You are comfortable navigating large, multi‑team codebases and resolving ambiguity where ownership boundaries meet. You combine strong low‑level understanding of GPU software with a practical appreciation for how AI frameworks are built, validated, and shipped. You communicate clearly across disciplines and earn trust through technical depth, consistency, and follow‑through. You naturally identify gaps between teams, turn them into concrete technical plans, and drive alignment without relying on formal authority.

Requirements

Deep experience with GPU software stacks, including libraries, runtimes, and performance‑critical code paths. Hands‑on familiarity with ROCm and its core components (HIP, Composable Kernel, MIOpen, hipDNN) or equivalent GPU platforms.
Practical experience working with AI frameworks such as PyTorch, Triton, or JAX, including operator integration or backend enablement. Proven ability to debug and resolve cross‑layer issues spanning libraries, compilers, runtimes, and frameworks.
Strong programming experience in C/C++ and Python, with comfort reading and reasoning complex codebases. Experience defining integration standards, performance criteria, and validation strategies across teams.
Demonstrated technical leadership in large, cross‑functional engineering environments.

Nice To Haves

Advanced degree in Computer Science, Computer Engineering, Electrical Engineering, or equivalent practical experience.
Advanced degrees, such as M.Sc., M.Eng., Ph.D. are preferred

Responsibilities

ROCm & Framework Integration Leadership: Define and own integration strategies between ROCm libraries (HIP, CK, MIOpen) and AI frameworks. Serve as the primary technical point of contact for framework teams on library behavior, performance characteristics, and integration constraints.
Performance & Quality Standards: Establish performance baselines, regression criteria, and validation expectations for framework‑facing library functionality. Partner with performance, tooling, and QA teams to ensure issues are detected early and addressed at the correct layer.
Cross‑Team Technical Alignment: Drive alignment across Libraries, Frameworks, Compiler, and DevTools teams on integration priorities and sequencing. Translate framework requirements into actionable library‑level work and provide clear technical feedback upstream.
Debugging & Escalation: Act as a senior technical escalation point for complex cross‑layer issues involving kernels, runtimes, compilers, and frameworks. Guide root‑cause analysis across organizational boundaries and ensure issues are resolved holistically.
Developer Experience & Adoption: Advocate for integration patterns that improve usability, debuggability, and out‑of‑the‑box performance for ROCm users. Capture and promote best practices through documentation, design reviews, and mentoring.
AI‑Assisted Engineering Practices: Leverage AI‑assisted development tools to accelerate analysis, integration validation, and documentation, while ensuring correctness and traceability.