AI VLA & SLAM Engineer

About Omakase Robotics

Omakase Robotics is building Omakase OS — a software and systems stack that turns robots into reliable workers for human spaces.

We started from software, but quickly learned that software alone is not enough for real-world deployment. Making robots useful in hospitals, hospitality, and retail requires hardware, autonomy, interaction, and operations designed together. Our focus is not lab demos — it's robots that work in the field.

We partner with leading platforms including Unitree and are actively validating robots in real environments such as hospitals. Our long-term goal: make robots practical, affordable, and widely deployable.

About this Role

As an AI VLA & SLAM Engineer, you will build the perception and spatial intelligence systems that let our robots navigate complex human spaces — hospitals, hotels, retail floors — not just warehouses. While competitors focus on warehouse AMRs and SLAM IP licensing, Omakase is deploying into unstructured, human-shared environments where the challenges are fundamentally harder and the impact is greater.

What You'll Do

• Research, develop, and deploy Vision-Language-Action (VLA) models for real-world robotic applications
• Design, optimize, and deploy 2D/3D SLAM algorithms to production robot platforms
• Build sensor fusion pipelines integrating LiDAR, cameras, IMUs, and wheel odometry
• Implement state estimation via EKF/UKF and graph-based optimization (g2o, Ceres)
• Validate models in simulation (Gazebo, Isaac Sim) and execute Sim-to-Real transfer on Unitree platforms
• Collaborate with the Data team on ELA (Embodied Learning from Action) training pipelines
• Integrate perception and SLAM systems with control, cloud, and hardware teams
• Track and apply latest research (NeRF, 3D Gaussian Splatting, VLA foundation models) to production

Required Qualifications

• 3+ years of professional or strong academic experience in VLA, SLAM, Visual Odometry, or VIO
• Production-level C++ (C++11/14/17) and solid Python skills
• Strong mathematical foundation: linear algebra, 3D geometry, non-linear optimization
• ROS/ROS2 development experience in Linux environments
• Experience processing real sensor data: 3D LiDAR, depth cameras, IMUs
• Deep learning experience with PyTorch or JAX

Nice to Have

• Experience with autonomous driving frameworks (Autoware, Apollo)
• Point cloud processing with PCL or Open3D
• GPU optimization with CUDA or edge deployment on NVIDIA Jetson/Orin
• Experience with ROS2 Navigation Stack, Behavior Trees, or BT.cpp
• Docker + CI/CD for robotics development workflows
• Proven Sim-to-Real track record on physical robots
• English-only candidates welcome — Japanese not required for this role

Who Will Thrive Here

• Thrive in an early-stage startup where you help define how things work
• Want direct, hands-on access to real robots — a level of freedom rare at larger companies
• Move fast, iterate quickly, and care about shipping things that work in the field
• Excited about Japan's first robotics OS platform and its real-world deployments (hospital, Tsukuba PoC)