Applied Scientist II, Advertising Trust
Posted
Amazon Advertising operates at a scale few places in the world can match, and every ad shown to a customer must meet a high bar for trust and quality. We are looking for an Applied Scientist II to advance the machine learning systems that evaluate advertising content at scale — balancing accuracy, latency, and cost so that we protect the customer experience without slowing advertisers down. You will own meaningful scientific problems end to end: framing the problem, building and validating models, and taking them to production where they make decisions on live traffic.
This is a hands-on science role with real ownership. You will work across large multilingual datasets, modern ML and large language model techniques, and tight engineering constraints, partnering closely with other scientists, engineers, and product teams to turn research into measurable customer impact.
Key job responsibilities
- Design, build, and ship ML models and systems that classify and evaluate content accurately and efficiently at scale.
- Translate ambiguous business and policy problems into well-scoped scientific problems with clear success metrics.
- Apply and adapt modern techniques — including deep learning, representation learning, and large/small language models — to production problems under real latency and cost constraints.
- Own the full model lifecycle: data, training, evaluation, deployment, monitoring, and retraining in response to drift and changing requirements.
- Define evaluation methodology and the metrics that matter (precision, recall, cost of error, latency), and defend trade-offs with evidence.
- Collaborate with engineering and product partners to integrate models into live systems, and mentor junior scientists.
This is a hands-on science role with real ownership. You will work across large multilingual datasets, modern ML and large language model techniques, and tight engineering constraints, partnering closely with other scientists, engineers, and product teams to turn research into measurable customer impact.
Key job responsibilities
- Design, build, and ship ML models and systems that classify and evaluate content accurately and efficiently at scale.
- Translate ambiguous business and policy problems into well-scoped scientific problems with clear success metrics.
- Apply and adapt modern techniques — including deep learning, representation learning, and large/small language models — to production problems under real latency and cost constraints.
- Own the full model lifecycle: data, training, evaluation, deployment, monitoring, and retraining in response to drift and changing requirements.
- Define evaluation methodology and the metrics that matter (precision, recall, cost of error, latency), and defend trade-offs with evidence.
- Collaborate with engineering and product partners to integrate models into live systems, and mentor junior scientists.