Pre-silicon Compute and ML Framework Manager

Every iPhone, iPad, Mac, Apple Watch, Vision Pro, and Apple TV runs on the compute and graphics software foundation built by Apple’s Graphics, Games, and Machine Learning team. The Pre-silicon Compute Frameworks team sits at the leading edge of that work designing the Metal and Metal compute frameworks and validating performance for features that will define future Apple Silicon GPUs.

We’re looking for an engineering manager with a strong technical background who’s excited to build and lead a team of engineers, and to partner with software and hardware leaders to shape Apple's GPU compute roadmap!

Description

As manager of the Pre-silicon Compute Frameworks team, you own the new APIs for Apple’s next GPU generations. You will lead, support, and grow a team of engineers building Metal and Metal compute framework features and conducting performance analysis on next generation GPU hardware, carrying features from emulation all the way through silicon and into the bring-up of new chips and products.

You will partner directly with leaders across Apple’s software and hardware organizations to propose, develop, and evaluate features for new silicon generations, representing Apple’s compute and ML roadmap while the hardware is still taking shape.

Minimum Qualifications

BS/MS/PhD in Computer Science, or a related field; or equivalent experience
2-3 years of experience as a manager or lead
Experience with at least one GPU compute API (Metal, CUDA, OpenCL, or similar)
Understanding of machine learning fundamentals, and familiarity with ML frameworks such as PyTorch, JAX, or CoreML
Knowledge of operating systems, computer architecture, and CPU/GPU programming, with experience in C/C++

Preferred Qualifications

Experience defining, planning, and executing complex projects involving partnership across multiple teams
Track record in strategic leadership and decision-making
Familiarity with Apple framework and API design patterns
Hands-on experience with ML frameworks such as PyTorch, JAX, or CoreML