Software Development Engineer - Machine Learning/AI Runtime

The Product Catalog is a strategic asset for Amazon. It powers unrivaled product discovery, informs customer buying decisions, offers a huge selection across a large number of categories and positions Amazon as the first stop for shopping online. Amazon Catalog Systems is looking for a customer-focused Software Development Engineer to help us make the world’s best product catalog even better and improve the experience for millions of customers. Amazon Catalog Systems leverages machine learning to find answers to the following questions from various unstructured data sources: - Does this TV have built-in WiFi and which streaming services are supported? - Can this wireless speaker play music from a flash drive? - Is the “Pack of 4 with 6 bottles of 8 fluid ounce” at $50 baby bottled formula cheaper than the “3 packs of 36 ounces” for $110 powder version? - Which cell-phone case is the most durable, ultra-thin and the best value for my money? - Is this mat made of silicone and what are its dimensions? - Where is this product manufactured? - When searching for “apple case” do you mean a cell-phone case compatible with an iPhone or a crate of apples? If you are excited about making the Amazon catalog more dynamic, smarter and changing the way we model and understand products and help customers discover, compare and purchase products, come join us! We are looking for people with initiative who enjoy diving deep into the data and coming up with innovative solutions. You will find challenges in: Scalability: We process billions of records about products ranging from electronics to cosmetics. We build highly distributed systems and design algorithms that are able to handle these large amounts of data and operate with sub second latency. Where traditional solutions fail we develop approximate, distributed and streaming algorithms. Ambiguity: We create, operationalize, monitor, and retrain/distill Large Language Models — running them at Amazon catalog scale, processing billions of requests per month. We build model-agnostic systems and platforms designed to not just survive but thrive in today's rapidly evolving AI landscape. As foundation models shift weekly and new capabilities emerge daily, our architecture adapts — and so do we. We're always raising the bar on managing ML and Gen-AI at scale. Responsibilities: - Partner with applied scientists and other engineers to take emerging agent and enrichment ideas from prototype to robust, instrumented production systems that can be evaluated, improved, and scaled. - Analyze and process large amounts of data to identify and extract valuable information from diverse sources (e.g., product catalog, customer reviews, search queries, product images, external knowledge bases) - Actively participate in idea and roadmap generation. - Work creatively through and around perceived limitations and/or challenges imposed by the delivery platform to enable delightful experiences for customers. - Effectively present work to all levels of the leadership. - Be an effective collaborator in a cross functional team of SDEs, Technical Program Managers, and Product Managers. Key job responsibilities * Design and build systems to extract structured knowledge from unstructured and semi-structured data sources that extend beyond existing catalog schema * Build LLM prompts and tune them to meet Precision and Recall requirements for novel attribute discovery * Build scalable, efficient, and automated knowledge discovery pipelines that surface new product dimensions for both conventional and agentic consumers * Analyze and process large amounts of data to identify and extract valuable information from diverse sources (e.g., product catalog, customer reviews, search queries, product images, external knowledge bases) * Define and evolve new schema patterns to represent previously uncaptured product knowledge * Actively participate in idea and roadmap generation Effectively present work to all levels of