AI Research Scientist (Multimodal Foundation Models)

Quick Facts

• Role: AI Research Scientist focused on multimodal foundation models (architecture, pre-training, and distillation)

Description

You will investigate how multimodal foundation models should be designed, trained, and distilled, including transferring capabilities from large models to smaller, efficient students. The work spans architecture, pre-training objectives, data strategies, optimization methods, and teacher–student learning techniques, with a major focus on models that run efficiently on Apple silicon and on-device. Deployment constraints (memory, latency, energy, privacy) are built into the research process from early experiments through final evaluation.

Responsibilities

• Develop and evaluate new multimodal model architectures, pre-training objectives, data strategies, and optimization methods

• Design distillation methods that preserve reasoning and multimodal understanding under constrained capacity

• Co-develop frontier and efficient models, including teacher–student training pipelines

• Incorporate compute, memory, latency, energy, and privacy constraints into architecture, pre-training, distillation, and evaluation

• Translate successful research into practical foundation-model technologies; may contribute to publications and open-source artifacts

Requirements

• Hands-on experience running large-scale pre-training experiments for large language models

• Experience with LLM pre-training topics: architecture, objectives, data mixtures, tokenization, curricula, scaling, and optimization

• Proficiency with modern deep learning frameworks (PyTorch or JAX) and distributed training systems

• Experience evaluating pre-trained models for language understanding, reasoning, instruction following, or multimodal capabilities

• Strong understanding of transformer-based architectures and efficient/scalable foundation-model training

• Master’s degree (or equivalent practical experience) in ML, CS, or related technical field

Benefits

• Opportunity to shape new research directions and run ambitious large-scale experiments

• Chance to translate research into product-impacting on-device intelligence

• Potential for publications and open sourcing of selected models, artifacts, evaluations, or tools