In this role, you will:
-
Design, develop, train and evaluate multi-sensor fusion based deep learning models to understand obstacles and environmental context
-
Understand and curate real and synthetic datasets to improve our models
-
Perform latency optimization and deploy models to our robot fleet
-
Build a deep understanding of Perception gaps and behavioral issues around difficult obstacle types in order to help plan and prioritize our work
-
Collaborate with Prediction/Planner team to deploy fully autonomous vehicles in environments with difficult and rare obstacles, extreme weather conditions, and complex driving scenarios
Qualifications:
-
5 years of industry experience or more
-
Proficiency in Python and some knowledge in C++
-
Deep Learning expertise, preferably with panoptic segmentation experience
-
Experience developing multi-sensor fusion algorithms for object detection, panoptic segmentation or object tracking
-
Familiar with Transformer architecture
Bonus Qualifications:
-
Technical leadership experience with software or machine learning teams
-
TensorRT or CUDA experience
-
Experience of 3DGS for 3D reconstruction or novel view synthesis