Machine Learning Scientist I - Performance Marketing

Booking Booking · Hospitality · Manchester, United Kingdom · ML Science

Machine Learning Scientist role focused on optimizing large-scale ML models for online bidding in performance marketing. The role involves developing advanced ML and optimization techniques for bidding algorithms, modeling user intent and marketplaces, and optimizing bidding strategies to maximize advertising budget efficiency. It also includes designing and implementing scalable evaluation pipelines, synthetic data generation, and benchmarking for model quality. The position requires a strong understanding of large-scale optimization, auction theory, and applying ML to industrial setups, with a focus on end-to-end research-to-production cycles and A/B testing.

What you'd actually do

  1. Develop innovative techniques for the next phase of our online bidding algorithms, including modeling user intent, modeling the online marketplaces, and optimizing our bidding strategy to maximize the efficiency of how we spend our advertising budgets.
  2. Design and implement scalable evaluation pipelines, including synthetic data generation and benchmarking for model quality, relevance, and consistency.
  3. Ensure the reliability, efficiency, and scalability of evaluation tools and frameworks in both offline and online environments.
  4. Conduct in-depth data analysis to define and track evaluation metrics, validate label quality, and explore performance across different traffic siloes.
  5. Collaborate closely with ML engineers to integrate evaluation components into production pipelines, supporting continuous improvement of bidding applications.

Skills

Required

  • Master’s degree or PhD required (Computer Science, Engineering, Mathematics, Artificial Intelligence, Physics)
  • Industry or academia knowledge of large scale optimisation techniques or mechanism design or auction theory.
  • Experience contributing to innovative machine learning and optimization solutions for large-scale business problems.
  • Relevant work or academic experience (MSc + 1 year of working experience), involved in the application of Machine Learning to business problems.
  • Knowledge of some machine learning facets: working with large data sets, model development, statistics, experimentation, data visualization, optimization, software development.
  • Understanding of cross-functional development of machine learning products (e.g. Developers, Commercial, Data Analytics, etc.).
  • Working knowledge of Python, SQL/BigQuery, Spark
  • Excellent English communication skills, both written and verbal.

Nice to have

  • Preferably evidenced by peer-reviewed publication, patents, open sourced code or the like.

What the JD emphasized

  • large-scale ML models
  • end-to-end research-to-production cycles
  • large-scale A/B testing
  • optimizing auction levers at scale
  • large-scale optimization techniques
  • auction theory
  • state of the art machine learning methodologies
  • scalable industrial setups
  • online bidding algorithms
  • modeling user intent
  • modeling the online marketplaces
  • optimizing our bidding strategy
  • scalable evaluation pipelines
  • synthetic data generation
  • benchmarking for model quality
  • reliability, efficiency, and scalability of evaluation tools and frameworks
  • offline and online environments
  • in-depth data analysis
  • define and track evaluation metrics
  • validate label quality
  • explore performance across different traffic siloes
  • integrate evaluation components into production pipelines
  • continuous improvement of bidding applications
  • align evaluation strategies with business goals and user impact

Other signals

  • large-scale ML models
  • end-to-end research-to-production cycles
  • large-scale A/B testing
  • optimizing auction levers at scale
  • large-scale optimization techniques
  • auction theory
  • state of the art machine learning methodologies
  • scalable industrial setups
  • online bidding algorithms
  • modeling user intent
  • modeling the online marketplaces
  • optimizing our bidding strategy
  • scalable evaluation pipelines
  • synthetic data generation
  • benchmarking for model quality
  • reliability, efficiency, and scalability of evaluation tools and frameworks
  • offline and online environments
  • in-depth data analysis
  • define and track evaluation metrics
  • validate label quality
  • explore performance across different traffic siloes
  • integrate evaluation components into production pipelines
  • continuous improvement of bidding applications
  • align evaluation strategies with business goals and user impact