Machine Learning Engineer, Foundation Model Services

Do you think differently? Are you eager to break the status quo, bold and ambitious, unafraid to take risks, and passionate about building best-in-class technology? If so, theres no better place to do it than Apple. The Foundation Model Services team builds the frameworks, services, and tools that run Apples largest foundation models in production. Our infrastructure powers intelligent experiences across products people use every day - Search, Music, TV, the App Store, Messages, Photos, Spotlight, Safari, Siri, and more - serving millions of queries at incredibly low latency while drawing every ounce of performance from our hardware. Join us and youll help bring intelligence to billions of users around the world, working on optimizing and serving large language, vision, and speech models at Apples scale. Work closely with product teams to build production-grade solutions that launch models serving customers in real time. Partner with foundation model researchers to prototype and develop inference for cutting-edge model architectures, and build tools that help us understand and remove performance bottlenecks across different hardware and use cases. Write high-quality code, learn quickly in a fast-moving field, and grow your impact as you take on larger pieces of the system.2+ years of industry experience building and shipping production software and/or machine learning systems. Proficiency in a modern programming language such as Go or Python. Experience deploying and operating services on a cloud platform (AWS, Azure, GCP, or equivalent) using containers and Kubernetes/Docker. 5 year+ industry experience in ML technologies (LLMs, Machine Learning, NLP, Information Retrieval, Statistics). Experience building or operating high-throughput, low-latency services. Strong communication and collaboration skills, with the ability to partner across research and product teams. Bachelor's degree or higher in Computer Science or related technical field.Familiarity with Nvidia TensorRT-LLM, vLLLM, DeepSpeed, Nvidia Triton Server etc.