Data Scientist, Algorithms - Lyft Ads

At Lyft, our purpose is to serve and connect. We aim to achieve this by cultivating a work environment where all team members belong and have the opportunity to thrive. 

Lyft Ads is one of Lyft’s newest and fastest-growing businesses, focused on building the world’s largest transportation media network. Our mission is to help brands reach riders during key moments of their journey—before, during, and after a ride—by delivering meaningful, contextually relevant ad experiences. We operate at the intersection of mobility data, real-time decision systems, and AI-powered personalization, enabling advertisers to run high-impact campaigns with measurable outcomes.

We are seeking an Algorithms Scientist  to help build the next generation of ads relevance, targeting, optimization, and measurement algorithms that power the Lyft Ads platform. In this role, you will work across large-scale datasets and complex real-time systems to design, prototype, and deploy production-grade machine learning models. You’ll collaborate closely with Engineering, Product, Data Science, and Sales to translate ambiguous business and advertiser needs into rigorous algorithmic solutions that improve ad performance, enhance marketplace efficiency, and drive meaningful revenue growth.

This is a high-impact, highly technical role within a rapidly scaling business line. The ideal candidate brings strong applied machine learning intuition, hands-on modeling experience, and the ability to write clean, efficient production code. You will play a critical role in shaping how advertisers connect with Lyft riders—pushing the boundaries of personalization, measurement, and real-time optimization in a dynamic marketplace.

• →Design, develop, and deploy production-grade machine learning models and algorithms that power core Lyft Ads capabilities, such as ad relevance, targeting, ranking, bid optimization, pacing, campaign delivery, and measurement.
• →Own the end-to-end lifecycle of modeling projects — including problem definition, data exploration, feature engineering, model development, offline evaluation, deployment, and monitoring.
• →Collaborate closely with Ads Engineering to integrate models into real-time ad-serving and batch decision systems, ensuring performance across latency, scalability, and reliability constraints.
• →Analyze large-scale mobility, behavioral, and ads performance datasets to identify patterns, surface opportunities, and guide ML and AI driven product improvements.
• →Implement rigorous model evaluation frameworks, including offline metrics, statistical tests, calibration, sensitivity analysis, and A/B experimentation to validate both model impact and system-level outcomes.
• →Build robust training pipelines, feature transformations, and scoring infrastructure, ensuring reproducibility, observability, and long-term maintainability.
• →Partner with Product, Engineering, and Sales to translate ambiguous advertiser goals (e.g., increased conversions, reach efficiency, brand lift) into measurable requirements and success metrics.
• →Investigate and resolve model behavior issues, production regressions, calibration drift, and performance anomalies in close partnership with Ads Infra teams.
• →Drive innovation by staying current with advances in ML for ranking, recommendation, causal inference, optimization, and ads measurement — and proactively identifying opportunities to apply them.
• →Contribute to Lyft Ads’ modeling and experimentation infrastructure, through model cards, documentation, reproducibility standards, and code quality improvements.

• Master’s, or PhD in Machine Learning, Computer Science, Statistics, Applied Mathematics, Engineering, or related quantitative fields; or equivalent applied industry experience.
• 3–5 years of hands-on ML/applied science experience, ideally involving production models, large-scale systems, or ads/recommendation/relevance domains.
  Strong proficiency in Python and machine learning frameworks such as PyTorch, TensorFlow, JAX, or scikit-learn; ability to write clean, efficient, production-adjacent code.
• Experience working with large-scale datasets and distributed data tools (Spark, Snowflake, Presto, Databricks).
• Practical experience building and evaluating:
• Ranking and relevance models
• Optimization or pacing algorithms
• Predictive models for CTR, CVR, or user response
• Causal or experimentation-based measurement methods
• Understanding of online/offline evaluation techniques, including:
• Offline metrics (AUC, NDCG, MRR, calibration)
• A/B testing methodologies
• Bias correction and counterfactual estimation
• Ability to solve ambiguous problems by structuring analyses, evaluating trade-offs, and proposing algorithmic solutions grounded in scientific rigor.
• Strong communication skills, with an ability to clearly explain model behavior, constraints, trade-offs, and recommendations to engineering, product, and sales partners.
• Demonstrated ownership of modeling work, including debugging, monitoring, documentation, and iteration after deployment.
• Curiosity, initiative, and a track record of delivering measurable improvements through high-quality modeling.