UnoMove hiring robot-brain engineers with vision-only AI, edge models

Younuo Zhixing (UnoMove), a vision-only embodied AI company based in Suzhou, China, is recruiting top talent to build the "brain architecture" for robots. Unlike competitors that rely on LiDAR stacking, UnoMove uses cameras + AI to achieve perception-decision-control in a single product called UnoBox. Code deploys on NVIDIA Jetson, algorithms train in Isaac Sim/Omniverse, and the final systems ship to real-world logistics yards, inspection robots, and agricultural machinery. The company is hiring for four senior roles: robot control & navigation engineers (PID/MPC/LQR, ROS/ROS2, SLAM), edge-side large model engineers (TensorRT, quantization, 3D Gaussian Splatting, NeRF, CUDA), full-stack robot application engineers (React/Three.js, WebSocket/MQTT), and likely embedded AI roles.

Key requirements include 4+ years of experience, C++/Python proficiency, hands-on work with real wheeled/legged robots or manipulators, and deployment experience on embedded platforms. For the model engineering role, UnoMove wants experts in PyTorch/TensorFlow, experience with Transformer-based multimodal models, and the ability to perform FP32→INT8/INT4 quantization. They emphasize Sim-to-Real transfer without middlemen — a lean, vision-only approach that minimizes sensor costs while maximizing AI leverage. The posting also highlights a startup culture with "unlimited snacks and coffee," appealing to engineers tired of corporate PowerPoint work. UnoMove is betting that camera-only perception combined with edge-deployed large models is the path to scalable, affordable embodied intelligence.