Principal Perception Engineer, Obstacle Foundation Models

What you'll do

• Define the technical vision, architecture, and roadmap for 3D obstacle perception in autonomous driving systems.
• Develop advanced 3D perception models using multi-camera inputs and multi-sensor fusion like radar and lidar.
• Build production-grade deep learning models using CNNs, transformers, large-scale pretraining, and parameter-efficient fine-tuning.
• Establish KPI frameworks to analyze real and synthetic datasets to improve model accuracy and robustness.
• Lead the data strategy by defining labeling requirements and implementing model-assisted workflows like active learning.
• Ensure perception solutions meet strict product requirements for safety, latency, and resource usage on embedded platforms.
• Provide technical leadership and mentorship to engineers across the broader perception and autonomy teams.

What we're looking for

• 15+ years of hands-on experience developing deep learning-based perception systems for complex real-world problems.
• Proven technical leadership as a senior or principal-level individual contributor owning features and architectural decisions.
• Proficiency in PyTorch and a track record of taking models from prototype to production.
• Strong programming skills in Python and/or C++ to build high-performance, production-quality software.
• Experience in data-driven development including collaboration on data strategy, labeling quality, and iterative model improvement.
• Expertise in 3D computer vision fundamentals, multi-sensor fusion, and transformer-based or BEV perception pipelines.
• Knowledge of advanced techniques like large-scale pretraining, distillation, LoRA, and vision-language models.
• BS/MS/PhD in Computer Science, Electrical Engineering, or a related field.