Apple AR/VR Job | AIML – Sr. Software Engineer, On-Device Machine Learning, Foundation Models

Job(岗位): AIML – Sr. Software Engineer, On-Device Machine Learning, Foundation Models

Citys(岗位城市): Santa Clara Valley (Cupertino), California, United States

Date(发布日期): 2023-7-28


Help us bring state-of-the-art foundation models to the phone in your pocket, enabling the next generation of ML-based experiences in a privacy-preserving way! Our team is responsible for the core framework that launches neural-network workloads on Apple devices. We build the bridge between the compute resources available on Apple hardware and an entire universe of ML models, trained by feature teams throughout Apple and by our developer community. Your work on our team will enable increasingly sophisticated models throughout our products, from the computer vision models that process every camera frame in the Apple Vision Pro, to potential large language models that could transform how we interact with our computing devices. By developing the underlying representation and pipeline for these workloads, and the mechanisms for mapping them to the CPU, GPU, and Neural Engine, you will play a critical role in expanding what is possible in the Apple ecosystem.


Excellent C/C++ programming and debugging skills

Passion for API design and software architecture

Outstanding verbal and written communication skills

Experience with modern neural-network architectures and deep learning libraries

Expertise with performance optimization (preferred)


– Design and implement improvements to Apple’s Model Intermediate Language (MIL), the intermediate representation of neural-network workloads shared across the inference stack

– Develop the mechanisms for analyzing and transforming MIL workloads

– Build the tightly integrated pipeline that optimizes and compiles models and then orchestrates their execution on device

– Collaborate with CPU, GPU, and Neural Engine hardware backends to push inference performance and efficiency

– Work closely with feature teams to facilitate and debug the integration of increasingly sophisticated models, including large language models