Principal Software Engineer (Perception Pretraining)

Join Waymo, a leader in autonomous driving technology. As a Principal Software Engineer in our Perception Organization, you will develop sensor-fusion foundation models for the Waymo Driver, optimize model training and deployment, and drive collaboration between teams. Enjoy a comprehensive benefits package, including medical, dental, and vision insurance, mental health support, competitive compensation, and a hybrid work model.

• Développer des modèles de fondation de fusion de capteurs pour le conducteur Waymo, capables de fournir une compréhension temporelle et spatiale de l'environnement.
• Résoudre l'entraînement à grande échelle et multi-étapes des modèles de fusion de capteurs en partenariat avec les équipes d'infrastructure de Waymo.
• Optimiser la latence de bout en bout de l'entraînement du modèle à son déploiement, et conduire des collaborations entre les équipes de perception.
• 6+ years of experience in a technical leadership role leading technical teams and setting technical directions in large ML Engineering organizations
• 3+ years of experience with computer vision models, or LLMs, or vision language models
• 6+ years of experience in ML-driven production systems that develops models with large-scale data, training, evaluation, and deployment
• Master's degree or PhD in Computer Science, Engineering, or a related technical field
• Proficiency in C++, Python, and modern deep learning toolkits like PyTorch or JAX
• 10+ years of experience in ML model development, and you have 4+ years experience with large-scale vision, video, or multi-modal foundation model development and their integration in end-to-end models
• Passion to lead engineering excellence and efficient model development through flywheel automation, while having direct technical experience in model quality development to achieve our goals
• Experience in multi-modal LLM pretraining and fine-tuning, and their infrastructure
• Experience in model distillation techniques and compute optimization
• Familiarity with end-to-end models and their development challenges