Staff ML Engineer – BEV & Multi-Modal Perception

Torc Robotics; Michigan; Hybrid; USD 216,000 - 259,000; Full time

Benefits offered by this job

Medical premiums for employees

401K with company match

Stock options

Paid vacation

Holiday closures

Life Insurance

Job summary

Torc Robotics is seeking a Staff Machine Learning Engineer to lead BEV-based perception models across perception tasks, unifying camera, LiDAR, radar data into coherent spatial representations. You will drive architecture, data pipelines, and distributed training at scale.

The role emphasizes technical leadership, model innovation, and collaboration with sensor calibration and mapping teams. Hybrid work in Ann Arbor, MI or remote in the US is supported.

Qualifications

• 10+ years of experience in deep learning for perception, 3D vision, and autonomous systems.
• MS or PhD in CS/EE/Robotics or equivalent experience.
• Proven BEV modeling, 3D scene understanding and multi-view fusion expertise.
• Strong Python and DL framework skills, e.g., PyTorch or TensorFlow.

Responsibilities

• Lead BEV model development and roadmaps for perception across tasks (detection, segmentation, road topology).
• Design advanced multi-modal architectures fusing camera, LiDAR, radar, and HD maps.
• Own large-scale training workflows and optimization.
• Mentor ML engineers and drive model innovation and validation.

Skills

BEV modeling

3D scene understanding

Multi-modal fusion

Python

PyTorch

Distributed training

Mentoring

Education

MS or PhD in CS/EE/Robotics

Tools

PyTorch

TensorFlow

Job description

Torc Robotics is seeking a Staff Machine Learning Engineer to lead BEV-based perception models across perception tasks, unifying camera, LiDAR, radar data into coherent spatial representations. You will drive architecture, data pipelines, and distributed training at scale.

The role emphasizes technical leadership, model innovation, and collaboration with sensor calibration and mapping teams. Hybrid work in Ann Arbor, MI or remote in the US is supported.