AI Research Scientist: Multimodal Foundation Models - Architecture, Pre-Training & Distillation

What you'll do

• Develop and evaluate architectures, pre-training objectives, data strategies, optimization methods, and teacher–student techniques.
• Co-design frontier and efficient models, incorporating memory, latency, energy, and privacy constraints throughout research and evaluation.
• Evaluate model capabilities and translate successful research into foundation-model technologies for Apple products.

What they ask for

• Hands-on experience designing, implementing, and running large-scale LLM pre-training experiments.
• Experience with LLM pre-training topics, including architecture, objectives, data mixtures, tokenization, curricula, scaling, and optimization.
• Proficiency with PyTorch or JAX and distributed training systems; experience evaluating pre-trained models.
• Strong understanding of transformer architectures and efficient or scalable foundation-model training.

Nice to have

• Experience contributing to major foundation-model pre-training efforts; relevant research contributions or experience with distillation and teacher–student training.
• Experience with multimodal models, inference efficiency, hardware–software co-design, or a strong research-impact record.

Pay

$205,400 and $308,500 (base salary); $216,200 and $324,800 (base salary)

Tools

PyTorch, JAX, Apple silicon

AI focus

Research and develop multimodal foundation models, including architecture, pre-training, scaling, optimization, distillation, evaluation, and efficient on-device deployment.