Principal Computer Vision Software Engineer

Magnendo Corp.•Newton, MA

15d

About The Position

Every minute matters in stroke treatment. Magnendo is developing robotic technology to make life-saving endovascular procedures faster, safer, and more accessible. By combining advanced robotics, magnetic navigation, and intelligent software, we are tackling one of the most technically demanding challenges in healthcare while aiming to expand access to advanced stroke care for patients around the world. Join us in shaping the future of endovascular intervention. We're looking for a Principal Software Engineer to lead the design and development of our perception and computer vision software spanning a wide variety of algorithmic techniques, sensing, and imaging technologies. These include both real-time and non-real time systems, 3D sensing and medical imaging, planning and real-time perception that drives robotic navigation and control.

Requirements

9+ years of experience in robotics, controls, and motion planning software engineering and demonstrated ability to deliver the types of software described above
A great collaborator who understands the importance of clear communication and humility
A software engineer who wants to build extensible, maintainable, high quality systems
Motivated to apply advanced technology across multiple fields to a concrete application that saves lives
Experienced with production software development in an environment with high reliability requirements, preferably under a regulatory regime
Strong fundamentals in robotic path planning, trajectory optimization, motion control, manipulator kinematics for redundant arms, and/or state estimation
A strong individual contributor first and foremost, with the interest and ability to grow into leading a small team as the group scales
Comfortable serving as a technical authority — providing design guidance and review to internal engineers and external contractors alike
Able to operate with the judgment and independence expected of a Principal-level engineer — sets technical direction with limited oversight, while holding the line on rigor where it matters most
Able to clearly convey complex robotics and controls tradeoffs to cross-functional teams and external partners
Proficient with Python, C++, Linux, GIT

Nice To Haves

Experience in the medical device industry, ideally with interventional or robotic systems
Experience with medical or surgical imaging (e.g., fluoroscopy, ultrasound, endoscopy, CT/MRI)
Experience with 3D sensing hardware (e.g., stereo cameras, structured light, depth sensors) and sensor fusion
Experience deploying ML models on embedded or real-time systems
Working knowledge of medical device software requirements (e.g., IEC 62304, IEC 60601)

Responsibilities

Own the architecture and technical direction of our computer vision software, including image processing, 3D sensing, and ML-based perception
Design and implement production software for computer vision and machine learning algorithms for feature detection, segmentation, registration, and continuous monitoring from medical imaging
Design and implement production software for 3D reconstruction using 3D sensors in a complex and variable environment
Take ML models from research and prototyping through validation and production deployment
Supervise and partner with third-party contractors to define requirements, validate deliverables, and develop strategies for integrating their contributions into our production system
Design a highly testable architecture with unit and integration tests developed in parallel software modules
Drive design decisions using first-principles engineering analysis, simulation, prototyping, and hands-on testing
Define and review requirements, test plans, and acceptance criteria for computer vision and ML software components
Collaborate closely with robotics, systems, and hardware engineers to integrate perception into the broader system
Support risk analysis, design reviews, and root-cause investigations for computer vision and ML software
Ensure that software is developed following a rigorous process that combines best practices of modern high reliability software development with a regulatory V-model framework