Senior Principal Machine Learning Engineer

About the role

You own the design and delivery of large-scale prediction and optimization for PubMatic Activate, shaping performance advertising outcomes across a diverse media landscape. This role demands deep mastery of machine learning fundamentals and robust production experience to directly influence key metrics such as CTR, CVR, VCR, CPA, and ROAS across dynamic environments like CTV, mobile app, and omnichannel campaigns. You will be responsible for architecting intelligent systems that learn from complex signals and drive measurable value for both publishers and advertisers. The position requires a strategic thinker who can navigate ambiguity while maintaining a rigorous focus on business impact and technical excellence. You will partner closely with cross-functional teams to ensure that models are not only accurate but also reliable and scalable in production. Your work will define the core intelligence that powers performance advertising decisions globally.

Key facts

What you'll do

Architect next-generation intake pipelines that unify auctions, impressions, clicks, video events, conversions, users, context, and inventory signals into a cohesive data foundation.

Design experiments that balance spend delivery, cost efficiency, campaign goals, marketplace dynamics, and system constraints to consistently deliver superior advertiser outcomes.

Craft and refine CTR, CVR, VCR, CPA, ROAS, app-install, user-value, and campaign-performance models using advanced ranking, calibration, and ensemble methods.

Redefine bidding strategies, pacing-aware optimization, exploration mechanisms, and value estimation to support performance advertising at massive scale and low latency.

Establish observation frameworks that close online/offline evaluation gaps, strengthen experimentation rigor, and harden production feedback loops for continuous improvement.

Remediate challenges such as sparse conversions, delayed feedback, biased logs, cold-start campaigns, attribution noise, and metric mismatch through principled modeling and careful data design.

Define model-ready features, labels, attribution windows, negative examples, training datasets, and serving requirements in collaboration with performance advertising signal partners.

Translate model outputs into real-time decisioning systems through tight collaboration with engineering, product, analytics, and platform teams to ensure seamless integration.

Mentor engineers and applied scientists while providing thought leadership on performance optimization strategies for ambiguous, high-impact problems across the media stack.

Elevate Activate from a media buying execution platform into a true performance optimization platform that delivers transparent, measurable value for partners worldwide.

Requirements

You must have hands-on experience building large-scale prediction or optimization systems for production environments that handle high throughput and low latency.

You need a strong grasp of supervised learning, ranking, calibration, causal thinking, experimentation, statistical evaluation, and model monitoring to ensure robust and fair outcomes.

You are expected to demonstrate experience constructing large-scale prediction or optimization systems in production with a proven track record of deployment and maintenance.

You should show history with CTR/CVR prediction, conversion modeling, bid optimization, value modeling, forecasting, calibration, or performance optimization in complex, data-driven systems.

You must possess the ability to reason about model quality, business impact, system constraints, production tradeoffs, and online performance to make balanced technical decisions.

You bring experience working with large-scale data and distributed ML workflows, including data preprocessing, feature engineering, and pipeline orchestration.

You wield strong engineering skills in Python, Java, SQL, Spark, TensorFlow, PyTorch, XGBoost, or similar technologies to implement scalable and maintainable solutions.

You thrive when providing technical leadership across ambiguous, high-impact optimization challenges, driving clarity and progress without direct supervision.

You hold a BS, MS, or PhD in Computer Science, Machine Learning, Statistics, Mathematics, Engineering, or a related technical field with a strong foundation in quantitative methods.

Nice to have

Experience in ads, search, recommendations, marketplaces, e-commerce, fintech, pricing, bidding, or real-time optimization systems that involve complex decision-making under uncertainty.

Familiarity with performance advertising goals such as CTR, VCR, CPC, CPA, ROAS, app install, retargeting, or user-value optimization across diverse channels and devices.

Knowledge of real-time bidding, programmatic advertising, ad serving, attribution, pacing, identity, incrementality, or performance advertising frameworks that connect data, models, and execution.

Practical notes

Hybrid engagement for the right candidate; location includes Redwood City, US, and Remote.

Compensation range is 180000 to 230000.

No specific hours, travel, visa, or application deadlines are stated in the source.