Deep Learning Library GPU Software Development Architect - Performance AI Libraries

About the position

The Deep Learning Library GPU Software Development Architect at AMD is responsible for architecting and optimizing deep learning and high-performance computing GPU kernels on the AMD ROCm platform. This role involves performance optimization of critical AI operators, guiding algorithm development for new GPU hardware, and collaborating across teams to deliver software solutions. The position requires a strong background in machine learning and deep learning, particularly in training and inference workloads, and emphasizes the importance of teamwork and communication in a fast-paced environment.

Responsibilities

• Architect and support AMD's Machine Learning and Deep Learning Library: MIOpen.
• Optimize deep learning training and inference workloads on massive parallel hardware and software platforms.
• Guide performance optimizing algorithms for new GPU hardware.
• Perform code reviews and build unit tests.
• Author detailed documentation related to work.
• Coordinate feature design and development across functional and organizational boundaries.
• Drive new hardware and software co-design to optimize deep learning workloads.
• Lead collaboration from hardware, compiler, to various deep learning frameworks.

Requirements

• PhD or Master's in Computer Science, Computer Engineering, or related fields, or equivalent experience.
• Strong programming skills in C/C++ with a focus on software architecture.
• Experience in performance optimization for GPU computing, particularly for deep learning.
• Knowledge of deep learning primitives, fusion, and inference optimization.
• Strong understanding of the software development lifecycle and practices including debugging, testing, and documentation.
• Ability to work independently and collaboratively within teams.

Nice-to-haves

• Experience with deep neural network machine learning technologies.
• Familiarity with modern machine learning programming frameworks.
• Experience developing virtualization containers and package managers for code deployment.

Benefits

• Employee stock purchase plan
• Annual bonus or sales incentive eligibility
• Competitive benefits package