Senior Machine Learning Perception Engineer - Fallback Driving System

Job Description

At General Motors, our product teams are redefining mobility. Through a human-centered design process, we create vehicles and experiences that are designed not just to be seen, but to be felt. We’re turning today’s impossible into tomorrow’s standard —from breakthrough hardware and battery systems to intuitive design, intelligent software, and next-generation safety and entertainment features.

Every day, our products move millions of people as we aim to make driving safer, smarter, and more connected, shaping the future of transportation on a global scale.

As a Senior Machine Learning Engineer on the State Estimation and Mapping (SEAM) organization, you will develop and improve the ML perception model that powers the secondary (fallback) autonomy stack for Super Cruise 3. You will focus on building robust perception from multi‑modal camera , lidar, and radar data so the vehicle can safely bring itself to a stop when the primary autonomy stack is unavailable.

You will lead the design, implementation, and continuous improvement of ML models for object detection, segmentation, tracking, and prediction, working closely with partner teams across perception , planning, controls, and safety.

What You'll Do

• Design, train, and evaluate ML perception models for object detection, semantic/instance segmentation, tracking, and short‑horizon prediction using multi‑modal camera, lidar, and radar data.
• Develop and maintain the secondary stack perception model that enables the fallback autonomy system to safely bring the vehicle to a minimal risk condition when the primary system experiences a fault.
• Define clear ML success metrics (e.g., precision/recall, latency, robustness under edge cases) and drive systematic experimentation to improve model performance against those metrics.
• Analyze large‑scale datasets, curate challenging scenarios, and build data selection and labeling strategies that improve robustness for long‑tail and degrad