Senior ML Engineer, Perception Research

About the role

This role focuses on advancing the capabilities of the Waymo Driver, the world's most experienced driver. Waymo, which began as the Google Self-Driving Car Project in 2009, aims to improve mobility access and save lives impacted by traffic incidents. The position is centered within the Waymo AI Foundations team, which develops machine learning solutions for autonomous driving challenges. The goal is to enable safe vehicle operations in numerous cities under all conditions. This hybrid role involves close collaboration with Principal Research Scientists.

The core mission involves creating machine learning solutions for open problems in autonomous driving. Current focus areas include reinforcement learning, learning from demonstration, generative modeling, Bayesian inference, hierarchical learning, and robust evaluation. You will work on projects that push the boundaries of perception and decision-making for self-driving technology.

You will architect multimodal perception pipelines that transform diverse sensor inputs into coherent world models. A key responsibility is owning experiments from initial data collection through final deployment, solving complex autonomy challenges. This requires meticulous analysis and seamless coordination between research and production teams. The role is integral to validating system readiness and ensuring stability before real-world implementation.

Your work will directly influence how autonomous vehicles perceive and interact with dynamic environments. You will tackle the challenges of processing sensor information from cameras, LiDAR, and radar to build reliable 3D perception. Success in this role demands a strong aptitude for rigorous experimentation and system-level thinking.

Key Responsibilities

You will design intake workflows that standardize heterogeneous sensor feeds for efficient downstream processing. This involves creating robust architectures capable of handling varied data formats and sources. You will also establish scalable build processes to support large-scale training regimes for world models and multimodal systems.

Designing review mechanisms to assess perception accuracy across different environmental conditions is a critical task. You will define evaluation criteria and methodologies to ensure model reliability. Orchestrating ship readiness validations will be a core function, ensuring models meet stringent stability and performance standards prior to release.

You will establish partner integrations to align model outputs with external ecosystem requirements. This involves collaborating with internal and external stakeholders to ensure seamless interoperability. A vital duty is calibrating data pipelines that manage terabyte-scale driving logs collected across varied geographies. You will implement optimization strategies to maximize throughput for both training and inference workloads.

Championing evaluation frameworks that quantify real-world performance of detectors and trackers is essential. This involves developing metrics and benchmarks to measure tangible outcomes. Your contributions will help maintain Waymo's position as a leader in autonomous driving technology.

Qualifications

Holding a PhD or Masters in Computer Science, Machine Learning, Robotics, or a similar technical field is required. You must bring at least two years of industry or post-doctoral research experience in Reinforcement Learning or Foundation Models. A deep understanding of distributed data parallel training using FSDP and other sharding approaches is necessary.

You must be capable of managing the complexity inherent to globally distributed inference infrastructure and deployment. This includes navigating challenges related to latency, synchronization, and system coordination. A willingness to engage with intricate production environments is essential for success.

Preferred Qualifications

Extensive experience training and deploying Computer Vision models for 3D perception tasks is highly valued. You should have a proven track record with production-quality 3D multimodal models and their respective encoders. Substantial involvement in high-impact industry AI projects with measurable outcomes is a significant advantage.

Experience in generative models for domains such as world models, images, videos, and 3D is beneficial. Techniques involving diffusion or autoregressive models are particularly relevant. These skills will help drive innovation in perception and prediction capabilities.

Compensation and Location

The expected base salary for this full-time position across US locations ranges from $153,250 to $232,000. Actual starting pay will be determined by factors such as work location, experience, relevant training, education, and skill level. This role is based in Mountain View, California, USA; New York City, New York, USA; Kirkland, Washington, USA; and San Francisco, California, USA.

Practical Information

This role follows a hybrid work schedule and reports to a Principal Research Scientist. Waymo provides comprehensive benefits to eligible US-based employees. These include health, dental, vision, life, and disability insurance. Retirement benefits include a 401(k) with company match. Paid time off consists of 20 days of vacation annually, accruing at a rate of 6.15 hours per pay period for the first five years of employment.

Additional benefits include 40 hours of sick time per year, with 5 days available per event at the company's discretion. Maternity leave (Short-Term Disability + Baby Bonding) offers 28-30 weeks. Baby Bonding Leave provides 18 weeks. The company observes 13 paid holidays per year. Confirmation of details is available What you'll do