Architect, Perception Data & ML Infrastructure

Requirements

• 10+ years of software engineering experience, with at least 5 years in a technical leadership role driving strategy for large-scale distributed systems or ML infrastructure
,
• System-of-Systems Architecture:
  Proven track record of architecting complex, multi-component platforms (e.g., connecting data ingestion, training pipelines, and evaluation loops) that serve 100+ internal engineers or millions of external users
,
• Expertise in Big Data & ML Ops:
  Deep, hands-on mastery of distributed data processing (Spark, Flume, Beam) combined with a strong understanding of ML life cycles (training, inference, embeddings, fine-tuning)
,
• C++ & Python Proficiency:
  Ability to read/write/debug complex C++ and Python code at a system level (e.g., optimizing memory usage in distributed jobs or designing high-performance C++ serving layers)
,
• Influence Without Authority:
  Demonstrated ability to align multiple Principals, Directors, and Staff engineers across different organizations (e.g., Infra vs. Product) toward a unified technical direction
,
• (Desirable) Foundation Model Infrastructure:
  Experience building the data infrastructure specifically for training Large Language Models (LLMs) or Vision-Language Models (VLMs) at scale
,
• (Desirable) Autonomous Vehicle Domain:
  Deep familiarity with sensor data (Lidar, Radar, Camera) and the unique challenges of robotics data (calibration, synchronization, latency)
,
• (Desirable) Active Learning & Data Flywheels:
  Hands-on experience accelerating & automating the learning process for at-scale ML learning systems
,
• (Desirable) Open Source Leadership:
  Significant contributions to major open-source data or ML projects (e.g., Apache Beam, Tensor Flow, PyTorch, Kubernetes)

• The Perception Data team at Waymo is responsible for the overarching strategy and technical steering of all data used to train and evaluate the Waymo Driver’s perception system
,
• We own the end-to-end data lifecycle, building the automated "flywheels" and "infra-as-product" solutions that transform millions of miles of driving sensor data into high-quality training sets
,
• Our work bridges the gap between raw data and advanced machine learning, focusing on complex challenges like active learning loops and open-vocabulary modeling
,
• By unifying data ingestion, curation, and evaluation into a seamless ecosystem, we enable the rapid development of foundation models and next-generation perception stacks
,
• We collaborate deeply across Machine Learning, Infrastructure, and Evaluation teams to solve "impossible" data problems, ensuring our models can reliably understand the long-tail of rare events
,
• Ultimately, our team provides the essential data foundation that allows the Waymo Driver to navigate the world safely
,
• In this hybrid role, you will report to a Director of Engineering
,
• Define Organizational Technical Strategy:
  Architect the 2-3 year Data vision for the entire Perception org, unifying the machine learning lifecycle into an automated & continuous flywheel
,
• Cross-Organizational Architecture:
  Drive high-stakes architectural decisions that span across Perception, Machine Learning, and Infrastructure organizations
,
• Technical Governance & Standards:
  Establish engineering excellence, API standards, and system reliability bars across the multiple teams under the Director (Data, Eval, Model Lifecycle), ensuring these distinct systems interoperate seamlessly
,
• Solve "Impossible" Data Problems:
  Lead the technical execution on the most ambiguous and complex challenges, such as designing & accelerating active learning loops that automatically curate & learn from rare long-tail events from millions of miles of driving data without human intervention
,
• Mentorship at Scale:
  Serve as a mentor to Staff and Senior engineers across the wider organization, growing the next generation of technical leaders and fostering a culture of rigorous design review