Amazon Science is building Ads Agent, an AI-powered agent that understands advertiser intent and executes across the Amazon Ads portfolio. They are seeking an experienced Applied Scientist to build intelligent agents that simplify advertising workflows and automate tasks for advertisers.
Responsibilities:
- Design, build, and evaluate agentic systems that plan multi-step workflows, invoke tools, and take autonomous actions across Amazon Ads products on behalf of advertisers
- Define evaluation frameworks and benchmarks for agent reliability, correctness, safety, and advertiser satisfaction
- Analyze agent behavior through deep data analysis and rigorous A/B experimentation to identify failure modes, measure effectiveness, and derive business insights
- Partner with engineers, product managers, and UX designers to ship end-to-end agent experiences that are scalable, efficient, and reliable at Amazon scale