Computer Vision Engineer

About the role

The Computer Vision Engineer at Mujin operates at the intersection of high-speed perception and industrial control within the dynamic Mujin ecosystem. You will join the Mujin Research and Development organization, concentrating on the creation and operational deployment of advanced computer vision solutions for demanding industrial settings. The primary mission involves enabling machines to recognize and determine the spatial orientation of diverse objects in challenging and variable environments. Your work directly supports the core capabilities of MujinOS and MujinController, powering autonomous functions that are critical in logistics and manufacturing automation. The role requires a unique balance of theoretical insight and practical engineering, demanding a strong design sense for managing complex three-dimensional spatial data. Simultaneously, you must adhere rigorously to production schedules and quality standards inherent to industrial deployment to ensure success. Ultimately, your contributions will be measured by the delivery of robust modules that function consistently and reliably on active factory floors under real-world conditions.

Key Facts

Position

Location: Japan

Work Arrangement: Full-Time

Language Requirement: English

Salary Range: 4000000 JPY per year

What you'll do

• Tackle complex scientific and technical problems related to recognizing and estimating the pose of a wide range of objects in difficult scenarios involving motion and variable conditions.
• Analyze and evaluate detection and pose estimation algorithms originating from academic papers and industrial research, refining these methods for real-world application.
• Integrate refined algorithms into a reliable production system using Python and C/C++ to meet strict operational specifications.
• Design, develop, and assess the performance of real-time computer vision applications that are highly scalable and accurate.
• Write efficient code in Python and C/C++ to implement solutions that meet stringent requirements for speed and precision.
• Conduct detailed testing and perform thorough performance analysis on algorithms to validate functionality and robustness.
• Utilize continuous integration tools essential for refining algorithms to support rapid deployment cycles and iterative improvement.
• Collaborate with hardware partners who supply custom cameras to ensure calibration data integrates correctly with runtime systems.
• Optimize build artifacts to ensure recognition latency remains low while accommodating a large catalog of unpredictable object types.
• Validate data paths from initial image capture through final storage to confirm that digital twin streams maintain consistency and remain traceable throughout the process.
• Guide the integration between vision processing modules and motion planning logic to maintain pose accuracy during high-speed operational cycles.
• Implement solutions that are adaptable to a wide variety of object categories and environmental conditions without sacrificing performance.
• Document technical processes and results clearly to facilitate knowledge sharing and future development efforts.
• Contribute to the improvement of existing pipelines to enhance reliability, efficiency, and overall system performance.

Requirements

• Possess a minimum of three years of professional experience building perception stacks for motion control in environments that are dynamic and constantly changing.
• Demonstrate exceptional fluency in Python and C/C++ to implement detection and pose estimation while adhering to strict latency constraints.
• Handle very broad categories of objects, necessitating adaptable models capable of generalizing across unknown and varied factory settings.
• Understand the challenges and constraints of industrial automation, including the need for systems that operate reliably 24/7.
• Have a strong grasp of computer vision fundamentals, including but not limited to feature detection, matching, and geometric reasoning.
• Be capable of working independently and as part of a collaborative team in a fast-paced research and development environment.
• Show attention to detail and a methodical approach to debugging complex vision pipelines that interact with hardware.
• Commit to following safety and quality standards required for deployment in active manufacturing and logistics facilities.

Nice to have

• Prior experience with digital twins and real-time systems in industrial automation is valued.
• Familiarity with MujinOS, MujinMI, and MujinController is considered an advantage for integration efficiency.
• Knowledge of camera calibration techniques and multi-sensor fusion methods.

Practical notes

Please verify all information on the official application page before proceeding.

The engagement for this role is Full-Time, and the position is based at the Mujin Headquarters in Tokyo, Japan. Candidates must be available to work in this location. The compensation for this role is set at 4,000,000 JPY per year. There are no additional travel requirements specified for this position, and the standard working hours apply. Visas may be required for international candidates, and the company will provide support in this regard. Please ensure you meet the eligibility criteria before applying.