Software Development Engineer (GPU Machine Learning Performance)

Responsibilities THE ROLE:

AMD’s Advanced Technology Group is an entrepreneurial research and development team to build AMD's future advanced platforms & products. Our teams work closely with outside customers and internal teams to develop hardware, software, and systems solutions into next generation computing platforms. As part of AMD Advanced Technology Group, you will have the opportunity to be part of a winning team that will collaborate with internal teams & customers, explore new platform technologies, and lead the development of best-in-class hardware, software, and systems technologies that our customers will use for real-world problems.

In this role, you will join our team working on AMD’s next generation GPU edge inference solutions, ensuring machine learning solutions hit performance goals through GPU performance analysis, profiling, workload optimization and other duties to unlock powerful experiences for end-users with AMD devices.

THE PERSON:

AMD is looking for an energetic, enthusiastic individual to join our Advanced Technology Group GPU based Machine Learning development to work on AMD’s long-term Graphics and Machine Learning solutions, which will have significant impact for AMD, and the future of our products.

Our group works on forward-looking projects and novel system concepts that significantly impact AMD’s future product portfolio. Your role will contribute to shaping AMD’s forward-looking hardware and software technologies.

KEY RESPONSIBILITIES:

• Work within a team on Machine Learning SW and Workload development in pre and post-silicon phases of product development enabling next-generation of ML workloads in functional and performance attainment
• Implement and maintain machine learning kernels, fusions, and optimizations
• Build clean, testable machine learning kernel and other code, contribute to microbenchmarks, and regression test suites.
• Tune parameters (tile sizes, block configs, memory layouts) to improve latency, throughput, and power.
• Contribute to runtime integration, production deployment, and tuning in various contexts such as inference frameworks, pre-silicon platforms, and customer use cases
• Collaborate across kernel, runtime, and model teams; participate in code reviews and design discussions.
• Collaborate with inter-disciplinary teams to improve system‑level efficiency, end-to-end performance, and power.

 

PREFERRED EXPERIENCE:

• Experience in parallel programming or GPU computing (e.g., CUDA/HIP/OpenCL), GPU graphics programming (DirectX, Vulkan), or high‑performance C/C++.
• Familiarity with machine learning architectures, transformers, U-Nets, CNNs
• Solid understanding of linear algebra and numerical computing concepts.
• Experience using profilers, interpreting performance metrics (e.g. occupancy, bandwidth, FLOPs, roofline basics).
• Demonstrated proficiency in C++ and Python
• Track record in using AI tooling to accelerate engineering work
• Strong collaboration and communication skills.

ACADEMIC CREDENTIALS:

• Master / Bachelor of Computer Systems, Computer Science, Software Engineering, or similar degree.

Qualifications Benefits offered are described: AMD benefits at a glance.

This posting is for an existing vacancy.