Staff ML Engineer, Perception Research

The mission of the Waymo AI Foundations team is to develop machine learning solutions addressing open problems in autonomous driving, towards the goal of safely operating Waymo vehicles in dozens of cities and under all driving conditions. As part of our work, we also initiate and foster collaborations with other research teams in Alphabet. AI Foundations areas that we are currently focusing on include reinforcement learning, learning from demonstration, generative modeling, Bayesian inference, hierarchical learning, and robust evaluation.

This role follows a hybrid work schedule and reports to a Principal Research Scientist.

You will:

• Conduct comprehensive experimentation to train and deploy state-of-the-art Multimodal LLMs and World models to perform 3D Perception using sensor information from Camera, LiDAR and Radar..
• Partner effectively with engineering and research teams across Waymo to deploy new models, and implement efficient workflows for model development and continuous training on new front-filled data.
• Apply and develop techniques such as quantization, pruning, knowledge distillation, and efficient attention mechanisms.
• Develop and maintain scalable data pipelines for Training & Eval to process data from multiple sources.
• Design and implement evaluation frameworks for perception models.
• Develop infrastructure for large-scale model distillation and bulk-inference pipelines for teacher models.
• Experiment with different model partitioning and sharding strategies to improve scalability and efficiency.
• Build and maintain tools for performance analysis, profiling (e.g., xprof), and debugging of ML models.

• PhD or Masters in Computer Science, Machine Learning, Robotics, or a similar technical field, with 4+ years of industry or post-doc research experience in Reinforcement Learning or Foundation Models. 
• Proficiency in implementing model training flows in a scalable, distributed and performant manner such as Data parallel, FSDP and other sharding approaches. 
• Proficiency in JAX, Flax, and potentially TensorFlow/PyTorch.
• A willingness to work with complexity of globally distributed inference infrastructure.
• Hands on experience with optimizing the training and inference of Transformer architectures

Nice to have

• PhD in Computer Science, Machine Learning, or Robotics, with a research focus on Reinforcement Learning, Foundation Models, or Multi-Modal learning.
• Substantial involvement in and contributions to high impact industry AI projects.
• Experience in generative models for domains such as world models, images, videos, 3D, using techniques such as diffusion or autoregressive models.
• Experience contributing to frameworks and libraries that improve training speed and scalability (e.g., JAX, Gemax, XManager)