Amazon Senior Applied Scientist SB Sourcing Relevance 

United States, Washington, Seattle 

411992356

Key job responsibilities* Design and implement machine learning models for SB ad retrieval, semantic search, and contextual relevance to improve ad coverage, engagement, and match quality.* Partner with product and science teams to define and evolve shopper relevance for Sponsored Brands—a complex and multi-faceted concept involving brand identity, product context, creative type, and shopper intent.* Build scalable methods to measure, model, and predict relevance across different ad formats and placements.* Investigate how to best apply relevance predictions in ad selection and auction ranking to improve long-term shopper engagement and advertiser value.* Develop multi-objective optimization techniques to balance competing goals such as relevance, CTR, brand discoverability, and marketplace fairness.* Lead large-scale A/B testing and offline evaluations to validate model impact and guide roadmap decisions.* Work closely with engineers to productionize models and ensure performance in real-time serving environments.* Stay current with advances in information retrieval, representation learning, and relevance modeling, and translate research into applied innovations.* Mentor junior scientists and contribute to hiring and growing a high-performing science team.* Communicate technical findings and strategic insights clearly to stakeholders across science, engineering, and product.

- 6+ years of building machine learning models for business application experience
- PhD, or Master's degree and 6+ years of applied research experience
- Experience programming in Java, C++, Python or related language
- Experience applying machine learning to solve real-world problems in large-scale systems
- Strong understanding of digital advertising, search, or e-commerce ecosystems
- Familiarity with statistical hypothesis testing, A/B testing, and causal inference for robust experimental evaluation
- Ability to translate ambiguous business goals (e.g., improving ad relevance, increasing advertiser value) into well-defined model objectives and metrics

- Experience with large-scale retrieval, relevance modeling, or search ranking
- Familiarity with LLMs or GenAI, and interest in applying them to improve ad sourcing and shopper relevance
- Knowledge of multi-objective optimization (e.g., gradient descent, Pareto optimization)
- Strong background in machine learning, including deep learning and representation learning
- Experience with search technologies such as Lucene, Solr, or ElasticSearch, and interest in innovating at the intersection of classical IR and modern ML retrieval
- Demonstrated ability to lead ambiguous problem spaces and validate solutions through experimentation
- Research publications or open-source contributions in ML, IR, or NLP are a plus