Staff Software Engineer, Discover AI Transformation, Genai Personalization

Google Google · Big Tech · Mountain View, CA +1

Staff Software Engineer focused on integrating Generative AI into Google Discover's personalization feed. The role involves architecting LLM integration layers, owning prompt engineering, evaluation, observability, and leading model selection/fine-tuning. The goal is to evolve Discover into a conversational and co-creative personal companion, impacting billions of users.

What you'd actually do

  1. Architect the LLM integration layer for real-time content generation, summarization, and format adaptation tailored to user context within the feed.
  2. Own prompt engineering, evaluation harnesses, and LLM observability tooling to maintain output quality, accuracy, and tone consistency at production scale.
  3. Drive data-driven quality iterations by analyzing user behavior and product needs. Define evaluation metrics and measurement approaches through offline evals and live experiments.
  4. Lead model selection, fine-tuning strategies, and cost-efficiency decisions across foundation models, balancing capability, latency, and scalability.
  5. Build LLM first features and solutions to evolve Discover from a place where users passively scroll to a conversational, adaptive, and co-creative personal companion.

Skills

Required

  • Java, C/C++ or Python
  • software development
  • software design and architecture
  • ML design and ML infrastructure (e.g., model deployment, model evaluation, data processing, debugging, fine tuning)

Nice to have

  • Master's degree or PhD in Computer Science or related technical field
  • ML or Quality experience working on Recommendation systems
  • Experience building agentic flows
  • Experience with user modeling, recommender systems and personalization
  • Knowledge of statistical methods, with excellent mathematical skills
  • Ability to drive quality projects end-to-end from design to implementation to eventual launch

What the JD emphasized

  • production scale
  • evaluation harnesses
  • LLM observability tooling
  • offline evals
  • live experiments
  • model selection
  • fine-tuning strategies
  • cost-efficiency decisions
  • foundation models
  • balancing capability, latency, and scalability
  • conversational, adaptive, and co-creative personal companion

Other signals

  • LLM integration
  • GenAI personalization
  • production scale
  • conversational companion