About the job
As part of the Consumer Monetization Platform Engineering team, you will be the technical lead for building production-grade MLAI models and inference systems that power our measurement intelligence platform. You will design and implement causal measurement models (brand uplift, sales uplift), multi-touch attribution systems (path to conversion), and AI agent architectures that automate advertising intelligence at scale.
Responsibilities
Design, train, and deploy brand uplift models using causal inference techniques (difference-in-differences, synthetic control groups, propensity score matching) to measure advertising effectiveness for brand campaigns
Build sales uplift and ROAS measurement systems that connect ad exposure to downstream conversion and purchase events, enabling closed-loop attribution reporting for performance advertisers
Develop multi-touch attribution and path-to-conversion models using Markov chains, Shapley values, and deep learning approaches to accurately value impressions across the full consumer journey
Design and implement feature engineering pipelinesfrom raw event data through feature computation, storage, and real-time servingthat power all measurement and optimization models
Build and optimize model training pipelines using PyTorch, with experiment tracking, hyperparameter tuning, and automated retraining workflows on large-scale advertising datasets
Deploy models as low-latency inference services using Vertex AI, with Pydantic-based API contracts, model versioning, AB testing, and canary deployment patterns
Build agentic AI systems using LangChain, LangGraph, and Google Agent Development Kit (ADK) for autonomous advertising intelligenceincluding yield optimization agents, publisher intelligence tools, and measurement reporting agents
Design and implement knowledge graph-powered reasoning systems using GraphRAG architectures that enable AI agents to reason over structured advertising data, audience relationships, and campaign context
Develop contextual bandit and reinforcement learning agents for dynamic yield optimization, including floor pricing, header bidding configuration, and demand partner allocation
Build behavioral embedding models that transform raw user signals into dense vector representations for audience intelligence, lookalike modeling, and real-time targeting
Collaborate with data scientists, product managers, and platform engineers to translate business problems into ML solutions with measurable impact
Establish ML observability: model performance monitoring, drift detection, automated alerting, and continuous improvement loops for all production models
Lead technical design reviews and mentor team members on ML engineering best practices, model architecture decisions, and production deployment patterns
Qualifications
Minimum
BS with 7+ years of relevant industry experience, or M.S.Ph.D. in Computer Science, Statistics, Machine Learning, or a related quantitative field with 5+ years of relevant industry experience.
Strong foundations in machine learning, statistical modeling, causal inference, and experimental design
5+ years of experience building and deploying production ML systems (not just researchprototyping)from feature engineering and model training through inference serving and monitoring
Proficiency in Python with strong software engineering practices: PyTorch for model development, Pydantic for data validation and API contracts, and production-quality code with testing and CICD
Experience with LLM application development frameworks: LangChain, LangGraph, or equivalent agent orchestration frameworks for building multi-step AI workflows
Experience with cloud ML platforms, preferably Google Cloud (Vertex AI, BigQuery ML, Dataflow) for training, serving, and managing ML models at scale
Experience with feature engineering and feature store patterns for large-scale ML systems
Proficiency in SQL and experience working with petabyte-scale data warehouses (BigQuery, Spark, etc.)
Experience with deep learning architectures: embeddings, transformers, sequence models, or graph neural networks
Strong understanding of AB testing, uplift modeling, and causal inference methodologies
Self-driven, challenge-loving, detail-oriented, with excellent communication skills and the ability to translate complex ML concepts for cross-functional stakeholders
Preferred
Experience with Google Agent Development Kit (ADK) or similar agent-native development frameworks
Experience with knowledge graphs, graph databases (Neo4j, Spanner Graph), and GraphRAG architectures for structured reasoning
Experience with contextual bandits, reinforcement learning, or multi-armed bandit algorithms in production environments
Experience in ad tech, programmatic advertising, or publisher-side monetization (measurement, attribution, yield optimization)
Experience with causal ML methods: difference-in-differences, synthetic control, instrumental variables, propensity score methods
Experience with privacy-enhancing technologies, differential privacy, federated learning, or clean room computation
Experience with MLOps tooling: MLflow, Kubeflow, Weights & Biases, or Vertex AI Pipelines
Background in NLP, information retrieval, or recommendation systems
Experience with distributed training (Horovod, DeepSpeed, FSDP) for large-scale models