Every time a customer searches, browses or asks an AI assistant for help on Amazon, the experience depends on great product data. Lead the team that defines what "high-quality" means for every product in our catalog of billions, and builds the LLM and agentic systems that measure it. Amazon's Catalog Systems division is looking for an exceptional engineering manager to lead, architect and deliver Generative AI-powered solutions that measure and raise the quality of the world's largest product catalog. An accurate, information-rich catalog powers product discovery, helps customers make confident buying decisions and makes Amazon the first stop for shoppers. That's true whether customers shop through search and detail pages or through conversational assistants and AI agents. You'll be the engineering leader behind the single quality score used across Amazon for every product. Brands and sellers use it to fix their listings. Category leaders use it to set goals. Search, shopping experiences and AI assistants use it to decide which products and details to show customers. Your systems use LLMs and multimodal models to find defects in titles, attributes, bullet points, rich content, variations and images. They catch text that contradicts the product image, weigh each defect by how much it hurts the customer, and roll it all into one score people trust. There's no single right answer here, and that's what makes it interesting. Some days it's a smart heuristic or a classic ML model. Other days it's LLM-as-a-judge, multimodal reasoning or an autonomous agent that improves its own prompts. You'll do all of this at a scale few places can offer, and your work will shape how millions of customers discover and buy products worldwide. What you'll work on: - GenAI and LLMs: Run production LLM and multimodal pipelines at catalog scale. Benchmark models, tune prompts and fine-tune for high precision at low cost. - Agentic systems: Build AI agents that detect defects, find root causes, check their own accuracy and close the loop with selling partners. Bring agentic coding practices to how your team builds software. - Powering customer experiences: Supply trusted, machine-readable quality signals that search, detail pages, recommendations and AI shopping assistants use to decide which products and details customers see. - Scale and cost: Build batch and real-time, event-driven systems on AWS (Bedrock, SageMaker, EMR/Spark, ECS, Lambda). Keep inference and compute costs growing more slowly than the catalog. - Measuring the measurers: Track precision and recall for every detector, and use human-in-the-loop auditing and smart sampling to build ground truth. - Insights and guardrails: Give brands and sellers fast feedback and self-service insights, and set up AI guardrails that block bad data before it reaches the catalog. What you'll own: - The vision, strategy and roadmap for AI-driven product quality measurement - Hiring, mentoring and growing a high-performing team of engineers and tech leads - Partnerships with science, image/media, catalog and selling partner teams to keep raising the quality bar - Technical decisions that balance innovation, accuracy, cost and operational excellence If you want your work to shape what millions of customers see every day, however they choose to shop, let's talk. Key job responsibilities - Own the long-term vision and strategy for AI-driven product data quality measurement, and turn it into a roadmap your team can deliver. - Lead a team of software engineers, including senior engineers, to architect, build, test and deploy scalable batch and real-time systems with a constant focus on customers. - Drive delivery of LLM, multimodal and agentic solutions that detect product data defects, weigh them by customer impact and roll them up into a trusted quality score. - Partner with science teams to benchmark models, tune prompts and fine-tune, so quality signals stay accurate as costs come down. - Set and track precision, recall and cost metrics for every detection system, and use the results to keep improving. - Work with image/media, search, catalog and selling partner teams to define product quality and turn measurements into fixes customers can see, across search, detail pages and AI shopping assistants. - Make technical decisions that balance speed, accuracy, cost and operational excellence. Manage technical debt and own system reliability and on-call health. - Hire, mentor and develop engineers and future technical leaders, and keep the team performing at a high level. - Report on progress, quality, operations and system performance to senior leadership, and influence their technical direction and investment decisions. A day in the life Your main job is making sure the product quality score is accurate, trusted and gets better every week. Most mornings start with your team's dashboards. Are the detectors catching real defects? Did a new LLM prompt raise precision or just cost? Is the real-time pipeline healthy? From there, the day is a mix of people and problems. You might work through a design with your engineers on scaling a new multimodal check to billions of products. Then you might sit down with scientists to decide whether a fine-tuned model is ready for production. You'll also meet with partner teams in image, search and selling partner experience, who rely on your score and bring you the hard questions. For example, why does this product get a low grade? Or can you flag this defect before it goes live? Your customers are both inside and outside Amazon. Brands and sellers use your signals to fix their listings. Category leaders use them to set goals. Shopping experiences, from search to AI assistants, use them to decide what customers see. Some days go beyond the usual work. You might dig into a customer complaint that shows a gap in how quality is measured, try out an AI agent that improves its own prompts, or explain to senior leaders why a metric moved. And you'll always make time to coach your team and help your engineers grow. About the team We believe every customer should be able to shop on Amazon with confidence in the product information they see. Our team builds the systems that measure product data quality across billions of products, then turns them into action. Brands and sellers use our insights to improve their listings, category teams use them to set goals, and shopping experiences from search to AI assistants use them to decide what customers see. You'll work with engineers, scientists and partner teams in image, search and selling partner experience. We value trust over coverage, design for billions from day one, and believe customer stories reveal what metrics miss.