Applied Scientist, Amazon Canada Technology, Amazon Canada

Role overview

Science and model development

• Develop personalized multimodal thinking-LLM techniques for customer, search, query, ranking, and product reasoning across primary and secondary languages.
• Train LLMs using shopping, returns, and reasoning signals, including continued pretraining, reinforcement-learning reasoning traces, and instruction tuning.
• Analyze Canadian ranking-quality gaps and benchmark model staleness and architecture against other marketplaces.
• Formulate testable ranking hypotheses that connect a feature, architecture, or training change to predicted metric movement.

Evaluation and collaboration

• Run standardized offline model evaluations, validate feature distributions and in-locale coverage, and lead experiments against offline and online benchmarks.
• Lead error analysis across query segments, categories, and devices, identifying Canadian bilingual and thin-category failure modes.
• Own metric tradeoff analysis and acceptable regression boundaries in relation to Canadian business priorities.
• Partner with Core Search on hypotheses, training, and quality gates; classify feature-expansion effort by availability, pipeline work, locale training, or feasibility.

Basic qualifications

• At least 3 years of experience building models for business applications.
• Experience in computer science, computer engineering, machine learning, or a related field.
• Programming experience in Java, C++, Python, or a related language.
• Experience in at least one of algorithms and data structures, parsing, numerical optimization, data mining, parallel and distributed computing, or high-performance computing.

Preferred qualifications

• Unix/Linux experience is preferred.
• A PhD, or a master's degree with relevant CS, CE, ML, or related-field experience, is preferred.
• Experience with performance metrics and developing measures of progress against KPIs is preferred.

Compensation

• Annual base salary range is CAD 149,300–249,300 in Vancouver or Toronto; total compensation may also include sign-on payments and restricted stock units.