Server ML Frameworks Engineer
Cupertino, California, United States
Posted 1mo ago
Job Location
Cupertino, California, United States
Tech Stack
Remote Work Policy
On-site
Categories
Machine Learning Engineer
About the job
As a senior technical leader on Apple's Server ML Frameworks team, you will set the technical direction and software architecture for the frameworks that power Apple Intelligence at scale. You will partner with teams across Apple to propose, develop, and evaluate features in new Apple Intelligence models; collaborate closely with our hardware organizations to influence the design of next-generation silicon; and help define the compute and programming models for future server architectures. You will also grow and mentor a team of engineers, establish engineering best practices, and build the cross-functional relationships needed to deliver complex, multi-team initiatives.
Responsibilities
Adding optimized GPU compute kernels across Machine Learning, Image Processing, Linear Algebra and Computer Vision.
Defining and implementing APIs in Metal Performance Shaders.
Performing in-depth analysis, compiler and kernel level optimizations to ensure the best possible performance across hardware families.
Partnering with Platform Architecture teams to define Apple GPU's compute hardware roadmap. Working with hardware team to analyze performance on future silicon
Tune GPU-accelerated compute across products.
Minimum qualifications
BS/MS/PhD in Computer Science or a related field, or equivalent practical experience
3–5 years of experience as a manager or technical leader
Experience with GPU kernel development and optimization using compute programming models such as Metal, CUDA, or similar
Experience with distributed training or inference techniques
Strong system-level programming skills in C/C++, with a solid understanding of operating systems and computer architecture
Preferred qualifications
Track record in strategic leadership and decision-making
Familiarity with Apple framework and API design patterns
Experience with graph compilers such as CuTE, CuTile, Triton, OpenXLA or LLVM is a plus
Good understanding of LLM and Diffusion based model architectures
