Senior Machine Learning Engineer – Perception & Embodied AI

Responsibilities

• End-to-End Model Lifecycle: Own the design, training, validation, and deployment of deep learning models for core perception tasks such as:
• 3D Object Detection and Tracking (vehicles, pedestrians, cyclists).
• Real-time map detection of the drivable world (lanes, road boundaries, traffic signs).
• Multi-Modal Sensor Fusion (Camera, LiDAR, Radar).
• Production Pipeline: Build and scale the ML training infrastructure, including data mining and loading, multi-stage training and evaluation, to streamline model development.
• Performance Optimization: Improve model performance through data iterations, parameter tunings, training strategy and architecture updates to produce reliable models that meet and the strict real-time, low-latency requirements on the vehicle's embedded hardware.
• Model Debugging: Conduct rigorous, data-driven analysis to identify, debug, and resolve performance degradations and failures, specifically targeting long-tail and adversarial scenarios (e.g., adverse weather, sensor noise, occlusions).
• Metric Definition: Define and implement robust model-level metrics to aid model development.
• System Integration: Work closely with the Safety, Systems, and other engineering functions to integrate Perception outputs.
• Skills & Experience
• BS, MS or PhD in Computer Science, Machine Learning, Robotics, or a related quantitative field.
• 5+ years of professional experience with a focus on Computer Vision, Deep Learning, and Perception in a production environment.
• Deep hands-on experience with modern deep learning frameworks (e.g., PyTorch or TensorFlow) for training, experimentation, and debugging complex DNNs.
• Proven experience working with and fusing data from multiple sensor modalities (Camera, LiDAR, and/or Radar).
• Practical experience deploying and optimizing ML models for resource-constrained, real-time embedded systems.
• Demonstrated ability to drive model improvements through large-scale data analysis, error logging, and data curation.
• Bonus:
• Expertise with Transformer-based models for 3D detection, tracking, and scene understanding.