Senior ML Research Engineer

Apple•Cupertino, CA

About The Position

Apple's System Intelligence and Machine Learning (SIML) team is seeking an applied scientist to advance multimodal scene understanding and human-centric device interaction. The role focuses on developing multimodal large language models (LLMs) that perceive complex environments and understand human behaviors, preferences, and interactions. It spans the generative model lifecycle, including architecture design, model training, and deployment. The role involves architecting, training, and deploying multimodal scene-understanding technology using visual and audio data; building and demonstrating new user experiences with large foundation models; and partnering with cross-functional teams in a fast-paced, iterative environment. Work includes advancing modeling across system architecture and core ML algorithms, modeling user behaviors and preferences from device and environmental interactions, retrieving insights from historical context, processing interleaved streaming data, reasoning across temporal horizons, designing long-term adaptive memory systems, and learning semantically rich representations for downstream applications. The engineer will deliver production-integrable ML models and solutions, including APIs for production-ready models and algorithms incorporated into training infrastructure.

Requirements

PhD or MS in Computer Science, Engineering, or a related field, with a focus on machine learning.
Strong research skills, including first-author publications in top-tier ML conferences.
Expert-level knowledge of the state of the art in large autoregressive transformer models, multimodal encoders, and representation learning.
Experience with multimodal large language models (LLMs).
Strong Python programming skills and experience maintaining ML codebases grounded in software engineering principles.

Nice To Haves

PhD in Machine Learning.
Proven track record of deploying innovative ML technologies in production.
Familiarity with developing ML for resource-constrained devices.
Experience working with large, cross-functional, and diverse teams.

Responsibilities

Architect, train, and deploy multimodal scene-understanding technology based on visual and audio data.
Build and demonstrate user experiences using large foundation models and develop innovative capabilities.
Partner with cross-functional teams and iterate rapidly in a dynamic environment.
Advance human interaction and scene-understanding modeling from system-level architecture through core ML algorithms.
Model user behaviors and preferences using device and environmental interactions.
Retrieve nuanced insights from historical context.
Handle deeply interleaved streaming data and reason across diverse temporal horizons.
Design memory systems for long-term adaptation.
Learn semantically rich representations for open-ended downstream applications.
Deliver ML models and solutions that integrate into production pipelines, including production-ready model APIs and algorithms built into training infrastructure.