Applied Scientist, Amazon Canada Technology, Amazon Canada
Core
Develop and train reasoning LLMs, embedding models, and multi-objective optimization systems to power personalized search and ranking for Canadian customers.
Role type
Applied Scientist (LLM & Ranking Systems)
Builds
Reasoning LLMs, embedding models, cross-encoder rankers, and multi-objective optimization systems for Amazon.ca
Domain
E-commerce, Large Language Models, Search & Ranking
Deliverable
production ML models
Required skills
Large language model training, multi-modal reasoning, reinforcement learning, numerical optimization, distributed computing, algorithm design, statistical analysis, hypothesis formulation, error analysis, metric trade-off analysis
Preferred skills
Unix/Linux, performance metric development, bilingual data handling
Technologies
Java, C++, Python, LLM frameworks, ranking systems
Responsibilities
Develop personalized multi-modal thinking-LLM techniques for search and ranking; Train large language models on domain-adaptive data and shopping reasoning traces; Analyze ranking-quality gaps and benchmark models against global equivalents; Formulate falsifiable ranking hypotheses linking features to metric movement; Run standardized offline evaluation and lead experiments to drive improvements; Lead error analysis on failing query segments to surface locale-specific failure modes; Own metric-tradeoff analysis and acceptable-regression boundaries; Partner with Core Search on hypotheses, model training, and quality gates; Classify feature-expansion effort tiers based on locale knowledge and science judgment.