3D Computer Vision Engineer

Engagement: Project contract, October–December 2026, with the potential to extend further based on prototype success.
Professional Fee: Rp20-50 million per month, based on directly relevant experience
Commitment: Full-time
Location: Remote
Travel: Reimbursed separately
We are seeking a hands-on technical lead for a confidential applied R&D project involving the analysis, matching and alignment of partial 3D geometry from real-world scan data.
This is an architecture and implementation role. The successful candidate will personally develop the core geometry pipeline, establish a rigorous evaluation methodology and advise whether additional technical resources are required.
Further information about the application domain will be shared with shortlisted candidates. Proprietary data and materials will be provided under NDA.
Responsibilities
Assess the quality, scale, density, noise and completeness of point clouds and meshes
Design preprocessing and geometry-quality-control procedures
Develop candidate retrieval and ranking for partial 3D fragments
Handle cases where no valid match exists
Implement robust coarse-to-fine 6-DoF registration
Generate and compare multiple alignment hypotheses where appropriate
Develop physical-fit measurements covering contact, gaps, surface consistency, collision and penetration
Structure results for review, correction, acceptance or rejection by human experts
Design held-out evaluation covering top-k retrieval, pose accuracy, false positives, runtime and human correction
Deliver modular, tested, reproducible and documented code
Produce a realistic technical delivery plan and evidence-based recommendation on any additional resources required
Required Qualifications
Demonstrable experience with point clouds, meshes, 3D registration, shape matching or geometric search
Strong Python engineering capability
Practical experience with Open3D, PCL, CGAL, Trimesh, PyTorch3D or equivalent tools
Strong understanding of rigid transformations, surface normals, spatial search, robust estimation and registration methods
Experience working with partial, noisy or incomplete real-world geometry
Ability to design controlled experiments and evaluate algorithmic failure cases
Ability to lead technical delivery while remaining directly involved in implementation
Clear communication with technical and non-technical domain experts
Relevant backgrounds may include photogrammetry, robotics or SLAM, metrology, reverse engineering, industrial inspection, medical 3D imaging, geometric reconstruction or cultural-heritage digitisation.
Experience limited to 2D computer vision, object detection, generative AI, 3D modelling or visualisation without geometric processing and registration is not sufficient for this role.
Selection Process
Shortlisted candidates will be invited to:
Present one or two relevant projects involving point clouds, meshes, 3D registration, geometric matching, or related computer-vision work.
Participate in a 60–90-minute technical interview covering their proposed approach, likely failure cases, evaluation methods, and implementation decisions.
Work through a simplified technical scenario during the interview. This will be a live discussion.
Provide professional references where appropriate.
How to Apply
Please send:
CV or professional profile
One to three closely relevant projects
Your exact personal contribution to each project
Methods, data characteristics and evaluation metrics used
Code sample, repository, paper, demonstration or technical write-up
One significant 3D matching or registration failure you encountered and how you investigated it
Location and availability through December 2026
Expected monthly professional fee
Applications without evidence of relevant 3D geometry work will not be prioritised.

Work arrangement
Yes