About Moving Atoms
Moving Atoms turns LiDAR and SLAM scans of real buildings, plants and warehouses into separated, measured, editable models: a BIM model for Revit and IFC, and a SimReady world in USD where robots train and are tested before they reach the site. We deliver within 48 hours, and we are building the team that makes that instant. We are a small team in San Francisco, backed by Y Combinator and NVIDIA Inception.
What you will do
- Build the models that segment a registered point cloud into individual objects: walls, slabs, pumps, vessels, racks, pipe runs.
- Fit clean, watertight geometry to each object and measure how far it sits from the scan.
- Handle the real world: occlusion, clutter, moving people and forklifts, noisy SLAM, scans of 200 GB.
- Turn every customer delivery that needed a human fix into training data and a regression test.
What we look for
You do not need every one of these. Strength in most is enough.
- Strong deep learning for 3D: point transformers, sparse convolutions, implicit or Gaussian representations.
- Hands-on experience with real scan data (terrestrial LiDAR, mobile mapping, SLAM or photogrammetry).
- Solid geometry and numerical skills in Python and PyTorch; C++ or CUDA a plus.
- A habit of measuring results against ground truth before claiming them.
How we work
- On-site in San Francisco, five days a week. Small team, short loops, decisions made with the evidence on the table.
- You will talk to customers yourself: BIM managers, surveyors, plant engineers and robotics teams.
- Competitive salary and meaningful early equity. Health, dental and vision cover. Visa sponsorship considered.
How to apply
Email [email protected] with a note about what you have built and links to anything that runs. A CV is optional. We read everything and reply to everyone.