ML Engineer — Foundation Models Data & Datasets

Apple Inc.; Cupertino (CA); Hybrid; USD 150,000 - 278,000; Full time

Benefits offered by this job

Relocation assistance

Employee stock purchase plan

Tuition reimbursement

Comprehensive medical and dental cover

Job summary

Apple Inc. is seeking a Machine Learning Engineer focused on Data for Foundation Models in Cupertino.

You will design data pipelines, curate large-scale image/video datasets, and apply synthetic data generation techniques to improve diffusion and autoregressive models used in Apple products. The role requires strong ML fundamentals, hands-on research experience, and expertise with PyTorch/JAX and large-scale data frameworks.

Qualifications

• MS/PhD in Computer Science or related field
• 3+ years improving image generation or understanding models via data ablation
• Experience with ML frameworks (PyTorch/JAX) and large-scale data tools (Hadoop/Ray)
• Experience in academic or industrial research environment
• Excellent communication, bias-for-action, and clarity in ambiguity

Responsibilities

• Design data annotation, enrichment, curation and synthetic data pipelines for foundation models
• Build high-quality datasets for training diffusion or autoregressive models
• Apply and extend genAI tools to improve data quality aligned with Apple products
• Stay at the forefront of AI research and rapidly adopt impactful developments

Skills

ML fundamentals

Data ablation experiments

PyTorch/JAX

Hadoop/Ray

Academic/industrial research

Education

M.S./PhD in CS or related field

Tools

PyTorch

JAX

Hadoop

Ray

Job description

Apple Inc. is seeking a Machine Learning Engineer focused on Data for Foundation Models in Cupertino.

You will design data pipelines, curate large-scale image/video datasets, and apply synthetic data generation techniques to improve diffusion and autoregressive models used in Apple products. The role requires strong ML fundamentals, hands-on research experience, and expertise with PyTorch/JAX and large-scale data frameworks.