About Yodo Labs
We build spatial intelligence for the physical world, connecting images, video, 3D scans, and field records to real assets, locations, conditions, and risks. We've just closed our first round of funding and are building the team.
What you'll do
Turn our 4D site models into sim-ready twins that robots can be tested on before they go to a real site. Each site arrives with metric geometry, one ID per object, labels, and materials. You'll give every object a mesh and collision shape, estimate its mass and friction from its material and size, add joints where parts move, and keep a photoreal Gaussian splat layer for camera-based robots. The output is OpenUSD that loads in Isaac Sim or MuJoCo, and every real robot run is checked against sim to keep the gap small. Read papers, test whether the ideas hold up on our sites, and turn the promising ones into working pipelines. You'll work directly with the founders, your work may ship into the product, and strong results can be published.
What we look for
- A foundation in 3D vision, physics simulation, robotics, or computer graphics
- Comfort with messy real-world data and the ability to work independently in an early-stage environment
- Working proficiency in either English or Japanese is enough; our team operates in both
Nice to have
Rigid body and contact simulation, sim-to-real transfer, system identification; mesh reconstruction, articulated object estimation, 3D Gaussian Splatting, physical property estimation from images; robot learning, digital twins, industrial inspection.
You don't need
A PhD, Japanese, prior industry experience, or every skill above.
What you'll gain
Hands-on work building simulation for real industrial sites, with access to site scans, LiDAR data, and GPU resources, and a front-row seat as robots are evaluated on twins of the places they will actually work.
How to apply
Send a note to pan@yodolabs.jp with the subject line "Research Scientist Intern, Simulation (Your Name)." A short description of what you've worked on, along with a CV, GitHub, or paper if you have one, is all we need to start a conversation.