Drimva

Apple

CVML Engineer - 3D Reconstruction

$150K–$278K

Sunnyvale, California · Full-time · Posted Oct 23, 2026

Experience
Minimum of 3 years relevant industry experience.
Education
BS degree in Computer Vision, Machine Learning, Computer Graphics, or related fields; MS or PhD preferred.

About the job

Researches and ships 3D/4D reconstruction and computer vision algorithms for human-focused health and fitness features, including runtime algorithms optimized for Apple devices.

What you'll do

  • Research, adapt, and implement 3D/4D reconstruction algorithms and combine classical geometry with deep learning to create training data and runtime algorithms.
  • Collaborate across teams to define requirements and validation frameworks.
  • Deliver production-ready code optimized for Apple silicon and mobile devices.

What they ask for

  • BS degree in Computer Vision, Machine Learning, Computer Graphics, or a related field, and a minimum of 3 years of relevant industry experience.
  • Hands-on experience developing 3D and/or 4D computer vision algorithms and systems, or equivalent; deep understanding of multi-view geometry and 3D reconstruction pipelines.
  • Strong proficiency in Python and PyTorch, with solid software engineering practices.
  • Solid mathematical foundation in linear algebra, optimization, and geometry processing; experience with modern ML architectures and their applications to 3D problems.

Nice to have

  • MS or PhD in a related field; experience with human-centric 3D/4D reconstruction, datasets, and evaluation.
  • Background in differentiable optimization or geometric deep learning; mobile ML optimization; SLAM, bundle adjustment, photogrammetry, or C/C++ proficiency.

Pay

$150,400 and $277,600 (base pay range; USD)

Tools

Python, PyTorch, C/C++, SLAM, bundle adjustment, Apple silicon

AI focus

Develops modern machine learning and computer vision methods for 3D/4D human reconstruction, ground-truth generation, and on-device health and fitness applications.

Model-written summary · the original posting is authoritative

First seen Oct 24, 2026 · Last verified Nov 5, 2026 (UTC)