HomeTechnologyProductsCareersInvestor RelationsAbout Us

Careers

Careers at DEEPMIRROR

Join DEEPMIRROR and help build the perception, decision-making, and control technologies at the core of next-generation robots. You will work alongside a leading robotics R&D team to bring products from concept to production, serve customers in China and around the world, and turn frontier robotics research into real-world systems.

  • SLAM/VIO Research Scientist

    Responsibilities
    1. 1. Develop visual–inertial, multi-sensor SLAM solutions, including optimized implementations based on frameworks such as FAST-LIO2 and VINS-Fusion. Deliver real-time, sub-meter localization in dynamic environments to support robot navigation and obstacle avoidance.
    2. 2. Build learning-based feature-matching pipelines using models such as SuperPoint and R2D2 to replace conventional feature methods and improve robustness in low-texture scenes and changing lighting conditions.
    3. 3. Develop semantic SLAM systems that integrate lightweight detection and segmentation models such as YOLOv5s and SEEM for dynamic-object filtering and semantic map construction, including improved approaches based on DynaSLAM.
    4. 4. Implement neural implicit mapping and real-time dense reconstruction using 3D Gaussian Splatting and NeRF, with support for semantic scene editing and physical interaction.
    5. 5. Optimize stereo depth-estimation engines by improving stereo-matching algorithms such as PSMNet and RAFT-Stereo, particularly for depth drift in low-texture regions.
    6. 6. Explore the integration of vision-language-action (VLA) models with SLAM. Embed natural-language instructions into the semantic map layer to support high-level task planning and execution, such as “open the third door.”
    7. 7. Design end-to-end motion-generation frameworks that combine neural maps with reinforcement learning methods such as PPO and SAC to produce feasible trajectories for robot navigation and manipulation.
    8. 8. Develop spatiotemporal consistency modules that combine optical flow, including Lucas–Kanade methods, epipolar geometry constraints, and deep learning to remove dynamic features in real time, targeting accuracy above 95%.
    9. 9. Build multi-scale map-update mechanisms that detect long-term environmental changes and support incremental map updates.
    Qualifications
    1. 1. PhD preferred in robotics, computer vision, or a related field, with at least three years of SLAM algorithm R&D experience.
    2. 2. At least one paper on SLAM published at a leading conference such as ICRA, IROS, or CVPR, or a record as a core contributor to an open-source SLAM project. Please provide links to relevant publications or code.
    3. 3. Familiarity with frontier learning-based SLAM frameworks such as SuperGlue, DROID-SLAM, and GS-SLAM.
    4. 4. Strong knowledge of multimodal sensor fusion, including tightly coupled optimization, Kalman filtering, and graph optimization using frameworks such as g2o or Ceres.
    5. 5. Expertise in stereo depth estimation, including stereo calibration, disparity optimization, and CUDA acceleration for stereo matching.
    6. 6. Proficiency with PyTorch, Open3D, and ROS 2, with the ability to independently implement high-performance algorithms in C++17 and Python.
    Preferred Qualifications
    1. 1. Hands-on experience deploying neural implicit mapping, including 3D Gaussian Splatting or NeRF, in robotic systems.
    2. 2. Practical experience integrating VLA models with SLAM.
    3. 3. Experience leading production-grade optimization of SLAM algorithms for dynamic environments and deploying them in real robot products.
    4. 4. Experience building large-scale SLAM datasets or deploying SLAM algorithms in production.

  • Computer Vision Model Deployment Specialist

    Responsibilities
    1. 1. Deploy visual-perception models on embedded platforms including Rockchip and NVIDIA Jetson, delivering low-latency, high-accuracy inference.
    2. 2. Lead model quantization and pruning, including INT8 and FP16 optimization, operator fusion, and memory optimization to improve edge-inference efficiency by more than 50%.
    3. 3. Develop hardware-acceleration interfaces using TensorRT, RKNN, and related toolchains. Optimize CUDA-core and NPU scheduling to maximize the performance of Rockchip and NVIDIA platforms.
    4. 4. Collaborate on deployment solutions for vision-language-action (VLA) models that enable embodied robots to perceive, interact with, and make decisions in real time.
    5. 5. Build end-to-end performance-evaluation systems covering power analysis, frame-rate monitoring, and thermal testing to ensure stable module operation.
    6. 6. Develop automated model-conversion toolchains and deployment standards that enable rapid team iteration.
    Qualifications
    1. 1. Master’s degree or above in computer science, electronic engineering, robotics, or a related field, with at least three years of experience deploying deep-learning models.
    2. 2. Expertise in the Rockchip and NVIDIA Jetson development ecosystems, including TensorRT, DeepStream, and RKNN-Toolkit2.
    3. 3. Strong command of PyTorch model-optimization techniques, including quantization, distillation, and ONNX export.
    4. 4. Proficiency in multithreaded C++ and Python programming, with experience in embedded Linux development and driver debugging.
    5. 5. A strong problem-solving mindset and the ability to independently resolve complex issues such as memory overflow and operator incompatibility.
    6. 6. Strong cross-functional collaboration skills and the ability to work efficiently with algorithm researchers and hardware engineers.
    Preferred Qualifications
    1. 1. Experience deploying robotic-vision modules, such as RGB-D cameras or camera–LiDAR fusion systems.
    2. 2. Familiarity with ROS 2 and simulation tools such as Isaac Sim and MuJoCo.
    3. 3. Publications on deployment optimization at leading conferences or meaningful contributions to open-source, high-performance inference frameworks.

  • Robotics Data Acquisition Systems Engineer

    Responsibilities
    1. 1. Lead the end-to-end delivery of multimodal human-behavior data acquisition systems. Design, develop, and deploy systems that integrate VR devices, motion-capture hardware, and robot sensors for vision, audio, force, and other modalities.
    2. 2. Build data acquisition platforms purpose-designed for robotics, enabling synchronized acquisition and management of visual, audio, motion, force, and other multimodal data.
    3. 3. Establish standardized data acquisition workflows, including experiment design, annotation guidelines, data-cleaning rules, and storage architectures, to ensure data quality.
    4. 4. Develop efficient acquisition and data-management toolchains that ensure integrity, consistency, and traceability throughout the data lifecycle.
    5. 5. Work with algorithm teams to build task-specific datasets for model training in areas such as human–robot interaction, intelligent control, and behavior recognition.
    6. 6. Track advances in robotic-sensing technologies, assess their practical value, and drive their adoption in real projects.
    Qualifications
    1. 1. Bachelor’s degree or above in computer science, automation, robotics, electronic engineering, or another relevant STEM field.
    2. 2. Familiarity with major robotic sensors, including RGB-D cameras, IMUs, and force or tactile sensors, plus hands-on experience integrating VR and motion-capture systems such as Vicon, OptiTrack, or HTC Vive.
    3. 3. Proficiency in at least one programming language—Python, C++, or C#—and project experience developing data acquisition interfaces or building experimental systems.
    4. 4. Knowledge of common approaches to data storage, annotation, and management, with foundational data-processing and analysis skills.
    5. 5. Strong communication and cross-functional collaboration skills, with the ability to independently drive data acquisition tasks from planning through delivery.
    6. 6. Project experience in robot data acquisition, VR experiment design, or human–robot interaction is preferred.
    Preferred Qualifications
    1. 1. Experience building and managing large-scale datasets.
    2. 2. Familiarity with ROS or ROS 2 and practical experience integrating robotic systems.
    3. 3. Research-project experience, publications, or patents in a relevant field.

  • ROS Middleware Engineer

    Responsibilities
    1. 1. Lead the architecture, core-module development, and lifecycle maintenance of ROS- and ROS 2-based robotics middleware, supporting coordinated operation across modules in automation systems and robot products.
    2. 2. Optimize and upgrade ROS and ROS 2 systems, focusing on communication latency, resource utilization, and stability. Improve DDS policies, node scheduling, and message-transport efficiency.
    3. 3. Work across algorithm, hardware, and software teams to develop custom toolchains and modules, including data synchronization, logging and monitoring, and fault-recovery components.
    4. 4. Build ROS monitoring and diagnostic platforms that track node communication and resource usage in real time, enabling rapid diagnosis of compatibility issues, packet loss, and other operational anomalies.
    5. 5. Help define standardized ROS interfaces and produce technical documentation and user guides to support team collaboration and downstream development.
    Qualifications
    1. 1. At least three years of ROS or ROS 2 development experience. Experience delivering service-robot or industrial-robotic-arm projects is preferred.
    2. 2. Strong C++ and Python skills, with proficiency in the rclcpp and rclpy frameworks and experience implementing topics, services, and actions.
    3. 3. Deep understanding of Linux, with proficiency in build tools such as CMake, Catkin, and Colcon, as well as development and deployment tools including Git and Docker.
    4. 4. Strong systems-analysis and problem-solving skills, with the ability to independently diagnose middleware bottlenecks and optimize communication performance.
    5. 5. Familiarity with robotic-system architecture and the ROS integration of sensor processing, motion control, navigation, and planning modules.
    6. 6. Strong communication and cross-functional collaboration skills, with the ability to drive projects and issues through closure.
    Preferred Qualifications
    1. 1. Practical experience with ROS 2 lifecycle nodes, QoS optimization, and custom message-interface design.
    2. 2. Familiarity with open-source DDS implementations such as Fast DDS and Cyclone DDS, or experience customizing them.
    3. 3. Experience integrating ROS middleware with multi-sensor fusion and SLAM algorithms.
    4. 4. Familiarity with Kubernetes and the deployment and operation of ROS systems in cloud-native environments.

  • Vision-Language-Action Algorithm Engineer

    Responsibilities
    1. 1. Lead end-to-end development of core vision-language-action (VLA) algorithms for robotics. Adapt and optimize multimodal models spanning vision, language, and action to create a closed loop from natural-language instruction and environmental perception to action execution.
    2. 2. Design and optimize multimodal model architectures. Starting from VLA baselines such as LLaVA and RT-2, develop lightweight approaches to cross-modal feature alignment and action-sequence generation for developer and enterprise use cases, improving real-time performance on resource-constrained hardware.
    3. 3. Validate algorithms in simulation and deploy them on real robots. Build Isaac Sim or PyBullet environments, deploy models to robot hardware including control boards and sensor modules, and address dynamic response, interference resilience, and other real-world challenges.
    4. 4. Improve task generalization and skill abstraction. Define atomic robot skills such as robotic-arm grasping and path planning, enabling VLA models to decompose and generalize complex tasks.
    5. 5. Track advances in VLA and embodied AI from leading conferences such as ICRA and CoRL. Bring techniques including pretrained-model fine-tuning and reinforcement learning from human feedback into products to continuously improve algorithm performance.
    6. 6. Support production and cross-functional delivery. Work with hardware and software teams to resolve deployment compatibility issues and produce technical documentation and deployment standards.
    Qualifications
    1. 1. At least three years of robotics or AI algorithm R&D experience. Practical experience applying VLA or multimodal foundation models to robots such as robotic arms or service robots is preferred.
    2. 2. Strong foundations in multimodal models, including familiarity with vision-language architectures such as CLIP, BLIP, and LLaVA. Proficiency with PyTorch or TensorFlow and experience with model fine-tuning, supervised fine-tuning, or reinforcement learning from human feedback.
    3. 3. Solid robotics fundamentals, including robot kinematics and dynamics, path planning with tools such as MoveIt or OMPL, and ROS or ROS 2. Ability to independently integrate algorithms with hardware.
    4. 4. Proficiency with robotics simulators such as Isaac Sim or PyBullet, strong C++ and Python engineering skills, and experience with model optimization and edge deployment using TensorRT or ONNX.
    5. 5. Strong foundations in mathematics and optimization, including convex optimization, numerical computation, and reinforcement-learning methods such as PPO and SAC. Ability to independently improve algorithm robustness in complex environments.
    6. 6. Strong cross-functional collaboration skills and the ability to work efficiently with hardware and software teams.
    Preferred Qualifications
    1. 1. Experience deploying VLA algorithms on humanoid robots or collaborative robotic arms, or developing robotics tools for external developers.
    2. 2. Familiarity with 3D vision, including point-cloud processing and 6D pose estimation, and perception technologies such as SLAM, with experience in multimodal fusion.
    3. 3. Publications on VLA or embodied AI at leading conferences such as ICRA, CoRL, or NeurIPS, or meaningful contributions to high-quality open-source projects.
    4. 4. Experience taking a VLA-based robotic product from zero to one and through production.

  • Robotics Test Engineer

    Responsibilities
    1. 1. Lead end-to-end testing for robotic systems, covering the integration of hardware, software, and algorithms. Define unit, system, and scenario-based test plans and cases across functionality, performance, and stability.
    2. 2. Test key modules, including the accuracy of cameras, IMUs, and LiDARs; algorithm functionality; and embedded-system stability. Identify and document issues accurately and track them through resolution.
    3. 3. Design and build automated robotics-test platforms and develop test scripts to improve efficiency and coverage and support rapid product iteration.
    4. 4. Work with R&D teams to analyze issues found during testing, provide complete test data, track corrective actions, and verify results.
    5. 5. Produce test reports and defect reports, maintain test cases and lessons learned, and continuously improve testing processes and standards to ensure delivery quality.
    Qualifications
    1. 1. Bachelor’s degree or above in computer science, electronic information engineering, automation, robotics, or a related field, with at least two years of experience testing robotic or embedded products.
    2. 2. Familiarity with robotic hardware and operating principles, including core modules such as vision systems, sensors, and motor control, plus knowledge of standard test methods and tools.
    3. 3. Proficiency in Python or shell scripting and Linux, with the ability to independently build test environments and use equipment such as oscilloscopes and logic analyzers.
    4. 4. Knowledge of ROS or ROS 2 and experience with robotic-system integration and scenario-based testing, with the ability to isolate hardware–software integration issues.
    5. 5. Strong logical reasoning and problem-analysis skills, a rigorous and responsible approach to work, and strong communication and cross-functional collaboration skills.
    Preferred Qualifications
    1. 1. Experience testing production AGVs, service robots, or industrial robots, with familiarity with robotics-industry testing standards.
    2. 2. Experience building automated test platforms, proficiency with Jenkins and Git, and knowledge of CI/CD workflows.
    3. 3. Knowledge of electromagnetic-compatibility and reliability testing processes, with relevant hands-on testing experience.
    4. 4. Experience optimizing testing processes and building systematic test-case frameworks.

  • Embedded Systems Engineer, Robotics

    Responsibilities
    1. 1. Lead end-to-end embedded-system development for robots using ARM, STM32, NVIDIA Jetson, and related platforms, including system bring-up, kernel customization, and software-architecture design.
    2. 2. Develop core drivers and control software for motors and sensors such as cameras, IMUs, and LiDARs. Implement PWM and PID control algorithms to ensure precise motion and stable sensor data.
    3. 3. Perform hardware–software integration and optimization with the hardware team. Resolve complex issues such as camera-image tearing and IMU drift while improving real-time performance and resource utilization.
    4. 4. Develop and integrate ROS 2 middleware. Build robotic communication frameworks and custom nodes that enable low-latency data exchange across perception, control, and decision-making modules.
    5. 5. Support production delivery by optimizing embedded-system manufacturing processes, resolving software issues during production, and producing technical documentation and SDKs for internal use.
    Qualifications
    1. 1. Bachelor’s degree or above in computer science, electronic engineering, automation, robotics, or a related field, with at least three years of embedded-systems development experience. Experience developing robot products is preferred.
    2. 2. Proficiency in C, C++, and Python, with experience in Linux or RTOS environments and a strong command of embedded-development and debugging workflows.
    3. 3. Familiarity with ARM architectures and mainstream embedded platforms, including STM32, RK3588, and NVIDIA Jetson, with the ability to configure device trees, customize kernels, and develop drivers.
    4. 4. Strong knowledge of ROS or ROS 2, including rclcpp and rclpy, with experience integrating multi-sensor fusion and robot-control modules.
    5. 5. Strong hardware fundamentals and the ability to diagnose low-level faults using oscilloscopes, logic analyzers, and related tools, plus strong cross-functional collaboration and problem-solving skills.
    Preferred Qualifications
    1. 1. Experience with joint calibration and data synchronization for camera–IMU–LiDAR systems.
    2. 2. Familiarity with real-time kernels such as PREEMPT_RT or Xenomai, or heterogeneous computing frameworks such as CUDA and OpenCL, with experience in system-performance optimization.
    3. 3. Experience leading embedded-system development through production for service robots, industrial robotic arms, or related products.
    4. 4. Familiarity with interfaces and protocols such as MIPI CSI, CAN, and Ethernet, plus experience designing reliable complex embedded systems.

  • Robotics Embedded Software Engineer

    Responsibilities
    1. 1. Port and customize Linux systems and kernels for RK3588/RK3566 and NVIDIA Jetson Orin/AGX Xavier platforms, and develop drivers for cameras, IMUs, LiDARs, and other hardware.
    2. 2. Improve system real-time performance and hardware utilization by more than 30% through kernel scheduling, interrupt handling, memory-management optimization, and related techniques.
    3. 3. Develop camera drivers using V4L2 and MIPI CSI interfaces, implement IMU data acquisition over SPI/I2C, and build synchronized pipelines for multi-sensor data.
    4. 4. Design sensor-calibration workflows and filtering algorithms, including Kalman filters, to ensure accurate and stable robot motion control.
    5. 5. Build ROS 2 communication architectures and optimize DDS performance for low-latency data exchange among perception, control, and decision-making modules.
    6. 6. Develop custom ROS 2 nodes that integrate core capabilities such as LiDAR SLAM, Nav2 navigation, and MoveIt 2 motion planning for robotic arms.
    7. 7. Lead bring-up and debugging for interfaces including MIPI, CSI-2, and GPIO, and resolve complex issues such as camera-image corruption and IMU drift.
    8. 8. Build power-consumption and real-time performance test platforms using tools such as perf, Ftrace, and oscilloscopes to validate operation in dynamic robotic environments.
    9. 9. Define driver-development standards and ROS 2 interfaces, and deliver production-grade SDKs that support model deployment and feature integration by algorithm teams.
    Qualifications
    1. 1. Bachelor’s degree or above in computer science, electronic engineering, robotics, or a related field, with at least three years of embedded-development experience.
    2. 2. Strong command of the Rockchip and NVIDIA development stacks, including RKNN-Toolkit2 and JetPack SDK, with experience in Linux kernel configuration and device-tree development.
    3. 3. Experience developing drivers for OmniVision or Sony IMX camera families.
    4. 4. Deep knowledge of the ROS 2 ecosystem, including lifecycle nodes and the rclcpp/rclpy frameworks, with the ability to optimize QoS policies independently.
    5. 5. Proficiency in C, C++, and Python, with experience writing shell scripts and building projects with Makefiles and CMake.
    6. 6. Strong low-level hardware diagnostic skills, including the ability to troubleshoot with logic analyzers and JTAG.
    7. 7. Ability to work effectively across algorithm and hardware teams and drive cross-functional projects to completion.
    Preferred Qualifications
    1. 1. Experience with multi-sensor fusion and joint calibration for camera, IMU, and LiDAR systems.
    2. 2. Familiarity with PREEMPT_RT or Xenomai real-time kernels, or experience with heterogeneous computing using CUDA or OpenCL.
    3. 3. Experience leading embedded-system development for production service robots or industrial robotic arms.

  • Senior C++ Software Engineer, Robotics

    Responsibilities
    1. 1. Lead the architecture, high-performance development, and iterative optimization of core modules for embodied robots. Develop low-latency, highly reliable C++ software for key real-world applications.
    2. 2. Contribute deeply to the integration and customization of robotics middleware, with a focus on ROS, ROS 2, and DDS. Optimize communication, node scheduling, and message transport to enable efficient coordination across modules.
    3. 3. Engineer key perception, decision-making, planning, and control algorithms. Integrate algorithm modules with hardware platforms, validate deployments, and ensure that functional and performance targets are met.
    4. 4. Build and continuously optimize low-level robotics software frameworks. Address challenges in real-time performance, stability, and resource utilization to improve overall system efficiency.
    5. 5. Follow team coding and quality standards and produce high-quality technical documentation, including architecture documents, API references, and developer guides.
    Qualifications
    1. 1. Bachelor’s degree or above in computer science, electronic engineering, automation, robotics, or a related field, with at least three years of C++ development experience. Experience in embodied AI, robotics, autonomous driving, or unmanned aerial vehicles is preferred.
    2. 2. Expert knowledge of C++ and modern C++ standards, including C++11, C++14, and C++17, with strong foundations in object-oriented programming, data structures, and algorithms.
    3. 3. Extensive hands-on experience with robotics middleware and a deep understanding of ROS and ROS 2, including node communication, messaging, and lifecycle management. Experience developing or architecting large-scale ROS projects is preferred.
    4. 4. Experience delivering projects in robotics, autonomous driving, or unmanned aerial systems. Familiarity with the complete perception–decision-making–planning–control software pipeline, hardware integration, and system optimization.
    5. 5. Strong technical problem-solving skills, with the ability to independently address complex system-development challenges and collaborate effectively across teams.
    Preferred Qualifications
    1. 1. Experience developing embodied robotic systems and integrating VLA models with robots.
    2. 2. Expertise in DDS implementations such as Fast DDS and Cyclone DDS, including middleware performance tuning or customization.
    3. 3. Familiarity with Linux systems programming and multithreaded or multiprocess development, plus experience with real-time systems such as PREEMPT_RT or heterogeneous computing using CUDA.
    4. 4. Experience with large-scale robotic-system deployment or productionization.

  • Senior Middleware Engineer

    Responsibilities
    1. 1. Lead the design, development, and lifecycle maintenance of cross-system middleware solutions that provide efficient and reliable communication across software components in complex applications.
    2. 2. Participate throughout the project lifecycle, from requirements analysis and technical selection to implementation, ensuring that middleware products meet both functional and system-performance requirements.
    3. 3. Analyze and remove performance bottlenecks in existing middleware. Improve stability, concurrency, and horizontal scalability to maintain availability under heavy workloads.
    4. 4. Work with frontend and backend teams to define middleware interfaces and development standards, participate in code reviews, and design unit tests to ensure delivery quality.
    5. 5. Create and maintain high-quality technical documentation, including design documents, interface manuals, and deployment guides, to support effective adoption and preserve institutional knowledge.
    Qualifications
    1. 1. At least five years of middleware-development experience, including the design or refactoring of middleware for complex distributed systems.
    2. 2. Strong command of C++ as the primary language, or Java or Go, with the ability to write high-performance, highly concurrent software.
    3. 3. Deep understanding of core middleware concepts such as message queues, service discovery, and configuration management, with the ability to design and implement highly available middleware solutions independently.
    4. 4. Strong analytical and troubleshooting skills, with the ability to diagnose and resolve complex issues during middleware development, deployment, and operation.
    5. 5. Excellent cross-functional communication and collaboration skills, including the ability to present technical solutions and enable other teams.
    6. 6. A strong interest in distributed systems and microservices, with a commitment to tracking advances and best practices in middleware.
    Preferred Qualifications
    1. 1. Experience customizing or making core contributions to open-source middleware such as RocketMQ, Nacos, or Dubbo.
    2. 2. Hands-on experience with middleware performance tuning, fault-injection testing, and disaster-recovery design.
    3. 3. Familiarity with Docker, Kubernetes, and middleware deployment and operations in cloud-native environments.
    4. 4. Experience building and operating middleware clusters for large-scale distributed systems.

  • Senior Hardware Engineer, Robotics

    Responsibilities
    1. 1. Lead end-to-end hardware design for core robotic systems, including schematic capture and PCB layout for main control boards, sensor modules, and motor-drive circuits, with a focus on reliability and manufacturability.
    2. 2. Select and validate key components based on product requirements, balancing reliability, performance, and cost. Design test plans to verify component suitability.
    3. 3. Independently bring up and debug hardware boards, perform functional testing and fault diagnosis, and work with software teams on hardware–software integration.
    4. 4. Lead electromagnetic compatibility and interference design and testing. Reduce interference through circuit layout, grounding, and other design measures, define EMC test plans, and drive corrective actions.
    5. 5. Support the full production lifecycle, including failure analysis, test-fixture design, manufacturing-process optimization, and resolution of hardware issues during production.
    Qualifications
    1. 1. At least three years of embedded-hardware development experience. Experience taking robotic products into mass production is preferred.
    2. 2. Expert knowledge of digital and analog circuit design, ARM architectures, mainstream microcontrollers, and peripheral-interface design.
    3. 3. Strong design capabilities for MOSFET gate-drive, isolation, and voltage-regulation circuits, plus extensive experience resolving EMC and EMI issues.
    4. 4. Proficiency with PCB design tools such as Altium Designer and Cadence, and laboratory tools including oscilloscopes, multimeters, and logic analyzers.
    5. 5. Strong cross-functional collaboration and problem-solving skills, with the ability to work efficiently with software and manufacturing teams.
    Preferred Qualifications
    1. 1. Experience integrating robotic sensors, including cameras, IMUs, and LiDARs, or designing motor-drive solutions.
    2. 2. Familiarity with power-integrity and signal-integrity analysis and optimization.
    3. 3. Experience with reliability and thermal design for complex hardware systems.
    4. 4. Participation in the complete development and production lifecycle of a robotic hardware product from zero to one.

  • Vision-Language-Action Algorithm Intern

    Responsibilities
    1. 1. Support data processing for VLA algorithms, including the acquisition, annotation, and preprocessing of images, point clouds, and text to build model-training datasets.
    2. 2. Assist with core algorithm development and validation, including VLA model fine-tuning and simulation testing. Record and analyze experimental results and help optimize model parameters.
    3. 3. Support algorithm deployment by developing and debugging ROS or ROS 2 nodes and assisting with integration testing in simulation and on real robots.
    4. 4. Track current research in VLA and embodied AI and prepare technical summaries and reports for the team’s R&D initiatives.
    5. 5. Help produce project documentation, including algorithm design documents and test reports.
    Qualifications
    1. 1. Currently pursuing a master’s degree or above in computer science, artificial intelligence, automation, robotics, or a related field, with strong foundations in deep learning.
    2. 2. Proficiency in Python, familiarity with PyTorch or TensorFlow, and basic experience training and debugging models.
    3. 3. Basic understanding of vision-language and vision-language-action models, including Transformers and imitation learning.
    4. 4. Strong learning ability and logical reasoning, with the ability to quickly learn robotics-development tools and workflows.
    5. 5. Responsible, detail-oriented, and collaborative, with clear communication skills. Available for at least three full working days per week for a minimum internship period of three months.
    Preferred Qualifications
    1. 1. Experience with ROS or ROS 2, or simulators such as Gazebo or MuJoCo.
    2. 2. Participation in university robotics competitions or research projects involving multimodal algorithms, data processing, or model tuning.
    3. 3. Familiarity with Linux and Git, with a foundational engineering mindset.
    4. 4. Ability to read technical English and a strong interest in VLA and embodied AI.

  • Linux Systems Engineer, Robotics

    Responsibilities
    1. 1. Customize Linux systems for robot hardware platforms, including system porting, kernel configuration, and scheduler optimization, to ensure real-time performance and stability.
    2. 2. Lead Linux driver development and debugging for cameras, IMUs, LiDARs, motors, and other peripherals, resolving compatibility and stability issues.
    3. 3. Design system-software architectures and interfaces, define standards for inter-module communication, and integrate application-layer software with low-level drivers.
    4. 4. Optimize system performance and resource utilization through memory management, interrupt optimization, and process scheduling to meet the demands of dynamic robotic applications.
    5. 5. Work with embedded and algorithm teams on hardware–software integration, provide system-level technical support, and document technical knowledge.
    Qualifications
    1. 1. Bachelor’s degree or above in computer science, electronic information engineering, automation, or a related field, with at least three years of embedded Linux systems-development experience.
    2. 2. Deep knowledge of the Linux kernel, including system porting, kernel configuration, device trees, and real-time optimization using PREEMPT_RT.
    3. 3. Proficiency in C and C++, Linux multithreading and network programming, shell scripting, and build systems including Makefiles and CMake.
    4. 4. Extensive driver-development experience and familiarity with interfaces including I2C, SPI, UART, CAN, and MIPI, with the ability to independently develop and debug peripheral drivers.
    5. 5. Strong cross-functional collaboration and problem-solving skills, with the ability to work effectively with hardware and algorithm teams. Robotics-development experience is preferred.
    Preferred Qualifications
    1. 1. Familiarity with ROS or ROS 2 and experience developing and integrating robotic-system software.
    2. 2. Experience developing on ARM platforms such as D-Robotics X5, RK3588, or NVIDIA Jetson, and familiarity with JetPack SDK and RKNN-Toolkit.
    3. 3. Experience optimizing Linux-system power consumption and designing for reliability, with hands-on delivery of production robotic products.
    4. 4. Familiarity with heterogeneous computing frameworks such as CUDA or OpenCL and experience with system-performance optimization and production deployment.