π Local Job Near You
Principal Applied Scientist
Microsoft Corporation
π
Redmond, United States
Location
Redmond
Posted
June 03, 2026
Commute
Local Area
Local Opportunity Near You!
This job is in your area. Enjoy a short commute and work close to home.
Job Description
**Overview**
Our Signals Modeling team builds the intelligence that powers how the advertising marketplace understands user behavior, measures impact and optimizes outcomes from initial impressions through downstream conversions and long-term advertiser value.
We develop large-scale learning systems that infer intent and causal effects from incomplete and noisy feedback, enabling principled decision-making across ranking, bidding, pricing, and budget allocation. Our work sits at the foundation of marketplace optimization, where accurate attribution and measurement directly influence billions in advertising spend.
The team designs and operates state-of-the-art modeling platforms spanning representation learning, weak-supervision, multi-objective training, calibration, and rigorous experimentation. We transform sparse engagement signals into reliable learning targets and build models that remain robust under delayed conversions, selection bias, and rapidly shif...
Our Signals Modeling team builds the intelligence that powers how the advertising marketplace understands user behavior, measures impact and optimizes outcomes from initial impressions through downstream conversions and long-term advertiser value.
We develop large-scale learning systems that infer intent and causal effects from incomplete and noisy feedback, enabling principled decision-making across ranking, bidding, pricing, and budget allocation. Our work sits at the foundation of marketplace optimization, where accurate attribution and measurement directly influence billions in advertising spend.
The team designs and operates state-of-the-art modeling platforms spanning representation learning, weak-supervision, multi-objective training, calibration, and rigorous experimentation. We transform sparse engagement signals into reliable learning targets and build models that remain robust under delayed conversions, selection bias, and rapidly shif...