3D Reconstruction Researcher (SfM & MVS)

Mountain View, California, United States expand job description ↓

Description

Geomagical Labs is a Bay Area AI startup, applying novel computer vision technologies to a mass-scale application to be used by millions.

We have an opening for a 3D computer vision scientist with strong academics and substantial experience in 3D reconstruction, applied mathematics, and applications of depth & semantics. As a 3D Reconstruction Researcher, you will work with our team of vision scientists in an applied R&D role, advancing the state of the art of 3D perception capabilities, for innovative mobile & cloud applications. This could be a dream job for highly-motivated applied researchers with entrepreneurial drive.

Candidates must have proficiency in sparse & dense 3D depth estimation (e.g. SfM, MVS, bundle adjustment, SLAM, direct methods), strong math skills in variational optimization & projective linear algebra, and have some exposure to image-based rendering techniques (e.g. novel view synthesis, mosaicing, super-resolution, synthetic focus). This position requires strong theoretical foundations, excellent software development skills, and sound scientific method. Experience using modern deep networks for perception is a plus.

Candidates must also be adventurous and excited about a seed-stage startup experience!

Requirements

  • Strong academic foundations from a competitive university/lab (Ph.D. preferred)
  • Strong problem solving and analytic skills; excellent scientific method
  • Expert in 3D geometric vision (e.g. stereo & multi-view geometry, camera pose & depth estimation, epipolar theory, warps, vanishing points, projective mathematics, ...)
  • Hands-on experience building 2D->3D reconstruction / photogrammetry technologies (SfM, bundle adjustment, MVS, direct methods, SLAM)
  • Substantial experience with energy-based non-linear and variational optimization for image-based applications
  • Strong applied math skills in linear algebra, PDEs, 2D/3D projective mathematics
  • Experience with 3D point cloud algorithms (e.g. ICP, EPnP, Kinect Fusion)
  • Strong C++ and/or Python software development skills in Linux environments
  • BONUS: Familiarity with image-based rendering and novel-view synthesis
  • BONUS: Experience implementing visual-inertial fusion for metric-scale odometry and/or reconstruction
  • BONUS: Experience implementing deep networks for perception (segmentation, depth, flow, recognition, etc.)
  • BONUS: Experience with 3D graphics & rendering
  • Must be adventurous and excited about an early-stage startup experience

Benefits

  • Rare opportunity to join an early startup adventure, led by founders who've taken two startup teams through IPO.
  • Substantial stock equity packages, of the sort only available to early startup employees.
  • Opportunity to publish novel and influential research and collaborate with partners in academia.
  • Strategic leadership roles possible for the right candidate.
  • Headquartered in Mountain View, California, in the heart of the Castro Street downtown, an easy walk from Caltrain and bicycling trails.
Personal information
Your Profile
Application Details
Do you have strong academic foundations in 3D vision from a competitive university/lab (Ph.D. preferred)?
Are you expert in 3D geometric vision, including stereo & multi-view geometry, camera pose & depth estimation, epipolar theory?
Do you have substantial hands-on experience building 3D vision reconstruction tech (SfM, bundle adjustment, MVS, direct methods)?
Do you have substantial hands-on experience with energy-based non-linear and variational optimization for image-based applications?
Have excellent analytical problem solving skills and experience analyzing, debugging, and improving algorithms using scientific method?
Have strong applied math skills in linear algebra, PDEs, 2D/3D projective mathematics?
Have hands-on experience with 3D point cloud algorithms (like ICP, EPnP, Kinect Fusion, mesh fitting)?
Strong MATLAB skills?
Strong C++ software development skills in Linux environments?
BONUS: Experience implementing visual-inertial fusion/SLAM for metric-scale odometry and/or reconstruction?
BONUS: Experience implementing deep networks for perception (depth estimation, segmentation, flow estimation, illumination estimation, etc.)?
BONUS: Are you familiar with the foundations of image-based rendering and novel-view synthesis?
BONUS: Experience with 3D graphics & rendering?
We are a very early startup company. Are you excited about seed-stage startups? Are you entrepreneurial? Are you comfortable with risk & uncertainty. Do you have the drive to make big opportunities from new ideas?