Machine Learning Engineer, Perception

About the role

This position is listed on behalf of a partner company, who manages all applications and next steps. Our partner is looking for a Machine Learning Engineer, Perception based in United States.

This is an opportunity to develop advanced AI and computer vision systems that enable industrial robots to perceive, understand, and interact reliably with the physical world. You will work at the intersection of machine learning, robotics, and welding science, building perception technologies that help automate complex manufacturing processes. Depending on your experience level, you may focus on implementing and optimizing models or lead the architecture and deployment of sophisticated perception systems. Your work will span 3D point cloud processing, deep learning, sensor fusion, and real-time robotic applications. You will collaborate with talented engineers to turn research into reliable, production-ready solutions that improve performance in real-world environments. Openings are available at experienced, senior, and staff levels, with opportunities for technical leadership and mentorship.

Accountabilities: -

Develop, implement, validate, and improve machine learning algorithms for weld perception, including point cloud registration, seam detection, and joint geometry estimation.

-

Expand perception capabilities across different weld joint types, part geometries, and industrial applications.

-

Build and maintain data pipelines for model training and evaluation, including annotated 3D scan ingestion, synthetic data generation, and structured dataset management.

-

Conduct rigorous model evaluation experiments, analyze failure modes, characterize false-positive and false-negative rates, and benchmark performance against quantitative accuracy targets.

-

Integrate trained models into production ROS-based robotics services, ensuring low-latency inference and compatibility with deployed robotic configurations.

-

Research and develop advanced perception algorithms using structured-light, RGB, and stereo sensors, including real-time in-process tracking and during-weld seam tracking.

-

Design and lead experiments involving advanced deep learning architectures, such as transformer-based models and geometric feature learning approaches.

-

Develop real-time perception systems using sensor fusion and probabilistic state estimation techniques, including Kalman filtering.

-

Own the end-to-end machine learning lifecycle, from dataset design and annotation strategies to model training, benchmarking, and deployment across robotic fleets.

-

Establish evaluation frameworks and go/no-go criteria, while improving data acquisition, annotation tooling, distributed training, and hyperparameter optimization workflows.

-

Assess the balance between synthetic and real-world scan data to improve model robustness and performance.