Senior Principal Data Scientist - Ads Measurement

It takes powerful technology to connect our brands and partners with an audience of hundreds of millions of people. Whether you're looking to write mobile app code, engineer the servers behind our massive ad tech stacks, or develop algorithms to help us process trillions of data points a day, what you do here will have a huge impact on our business-and the world.

We are an industry-leading direct-to-consumer and ad tech solution for advertisers and publishers. Our innovative ad tech gives one-stop access to Yahoo, Inc.'s trusted data, high-quality inventory and demand, creative ad experiences, and industry-leading machine learning at global scale.

As a Senior Principal Data Scientist on the Consumer Monetization Platform Engineering team, you will be the technical leader defining how machine learning and AI are applied to close the loop between ad serving and advertiser business outcomes. You will design and build the ML models, reinforcement learning systems, feature generation pipelines, experimentation frameworks, and feedback architectures that transform our platform from impression-based optimization to outcome-driven intelligence.

Our Big Data footprints are among the largest few in the world, at double-digit petabyte scale. You will work with hundreds of billions of ad events monthly, building models that learn from delayed conversion signals, sparse reward data, and complex multi-touch attribution paths. The challenges span real-time prediction at sub-100ms latency, offline reinforcement learning from logged bandit feedback, causal inference for experimentation, and multi-objective optimization balancing publisher yield with advertiser outcomes.

If you are someone who thrives at the intersection of rigorous ML research and production-scale engineering, who gets excited about building learning systems that improve autonomously from real-world feedback, and who wants to shape the science strategy for a platform generating billions in advertising revenue, we want to hear from you!

• Design and train production-grade ML models for conversion prediction, click-through rate estimation, engagement scoring, and advertiser ROAS forecasting at petabyte scale
• Develop and deploy multi-task and multi-objective learning models that jointly optimize for publisher yield and advertiser outcomes
• Build offline and online model training pipelines with automated retraining, model versioning, validation, and canary deployment workflows
• Apply advanced ML techniques including gradient-boosted trees, deep neural networks, transformer architectures, and ensemble methods to advertising optimization problems

• Design and build real-time and batch feature generation pipelines that capture user intent signals, contextual relevance, behavioral patterns, and advertiser performance indicators
• Develop feature stores and feature serving infrastructure that provides low-latency access to hundreds of features at prediction time
• Create novel features from cross-channel signals-search intent, content engagement, purchase behavior, and ad interaction history-to improve model accuracy
• Establish feature importance analysis, drift detection, and automated feature quality monitoring

• Design and lead the experimentation strategy for advertiser outcome optimization, including A/B tests, multi-armed bandit experiments, and interleaving designs
• Build and maintain the statistical framework for experiment analysis-power calculations, significance testing, sequential analysis, and correction for multiple comparisons
• Develop automated experiment monitoring, guardrail metrics, and early-stopping criteria to protect revenue while enabling rapid iteration
• Translate experiment results into actionable insights for product, engineering, and business stakeholders

• Design and build the closed-loop feedback architecture that connects ad delivery decisions to delayed conversion events, post-click engagement, and advertiser business outcomes
• Develop reinforcement learning (RL) and contextual bandit systems for real-time bid optimization, dynamic floor pricing, and ad ranking that learn continuously from outcome feedback
• Implement offline policy evaluation techniques (inverse propensity scoring, doubly robust estimation, replay methods) to safely evaluate new RL policies before online deployment
• Design reward shaping and credit assignment mechanisms that handle delayed rewards, sparse conversion signals, and multi-touch attribution across the ad delivery lifecycle
• Build autonomous learning systems where optimization agents self-improve from real-world feedback without manual intervention, with appropriate safety constraints and guardrails

• Define the ML and data science strategy for closed-loop measurement and advertiser outcome optimization across the monetization platform
• Mentor and provide technical…