Machine Learning Engineer, Search and Shopping

Build and scale low-latency machine learning for Search and Shopping Ads, optimizing predicted click-through rate (pCTR) models and ad integration in emerging AI search experiences.

Responsibilities

• Own technical architecture, delivery, and cross-team strategy for Search and Shopping Ads pCTR models in close partnership with DeepMind, Research, and Ads Machine Learning teams.
• Design, prototype, and scale high-capacity pCTR architectures that maximize Tensor Processing Unit (TPU) performance while meeting strict low-latency serving and return-on-investment budgets.
• Develop modeling solutions that capture deep user history and nuanced attention signals, integrating ads into emerging AI Search experiences including AI Overviews and AI Mode.
• Engineer mathematical loss functions and calibration methods to translate business objectives into improvements across top-line metrics and auction outcomes.
• Build agentic machine learning workflows to automate and accelerate optimal model architecture and feature space discovery.

Requirements

• Bachelor’s degree or equivalent practical experience.
• 8 years of experience with software development, including 5 years with large-scale machine learning, deep learning, neural networks, or recommendation systems.
• Experience designing and implementing large-scale production deep learning or neural network architectures under latency and computational constraints.
• Experience leading cross-functional technical projects and mentoring other engineers.

Preferred Qualifications

• PhD degree in Computer Science, Machine Learning, Artificial Intelligence, or a related field.
• Experience with agent-driven ML exploration, hyperparameter tuning, or automated model architecture search.
• Experience with one or more of the following:
  • Loss engineering for business objectives
  • Joint modeling across distinct prediction stacks
  • Hardware-aware ML optimizations (example: leveraging dense compute/TPUs effectively)
• Familiarity with ads prediction systems, auction dynamics, or serving infrastructure (example: AdBrain, Admixer).
• Ability to collaborate with peer technical leads and advanced ML research organizations (such as DeepMind or Google Research) to translate academic or exploratory methods into production systems.

Technologies

• Tensor Processing Unit (TPU)
• Large-scale machine learning
• Deep learning
• Neural networks
• Recommendation systems
• Sequence modeling
• Agentic artificial intelligence workflows
• Agentic machine learning workflows
• Loss functions
• Calibration methods

Compensation and Location

• Location: Mountain View, CA (onsite)
• Salary: USD 207,000 - 300,000 per year
• US compensation package: $207,000 - $300,000 (USD) + 20% bonus target + equity + benefits
• Learn more about benefits at Google.

About the Job

• Invent novel low-latency architectures that evaluate layouts in milliseconds while maximizing TPU capabilities.
• With DeepMind and Research, design sequence modeling to capture deep user history across AI experiences including Artificial Intelligence Overviews and Artificial Intelligence Mode.
• Engineer loss functions for auction dynamics and deploy agentic artificial intelligence workflows to accelerate model discovery.