Announcing the public release of Kaputt, a large-scale dataset for visual defect detection in retail logistics. With 238,421 high-resolution images of 48,376 unique items, Kaputt is 40 times as large as current state-of-the-art benchmark datasets and captures the real-world complexities of detecting defects across a vast range of products: https://amzn.to/4nAVCpV
Amazon Science
Research Services
Seattle, Washington 379,663 followers
The latest news and research from Amazon’s science community. #AmazonScience
About us
Amazon Science gives you insight into the company’s approach to customer-obsessed scientific innovation. Amazon fundamentally believes that scientific innovation is essential to being the most customer-centric company in the world. It’s the company’s ability to have an impact at scale that allows us to attract some of the brightest minds in artificial intelligence and related fields. Our scientists continue to publish, teach, and engage with the academic community, in addition to utilizing our working backwards method to enrich the way we live and work. Follow us on LinkedIn and visit our website to get a deep dive on innovation at Amazon, and explore the many ways you can engage with our scientific community. #AmazonScience
- Website
-
https://www.amazon.science
External link for Amazon Science
- Industry
- Research Services
- Company size
- 10,001+ employees
- Headquarters
- Seattle, Washington
- Founded
- 2020
- Specialties
- Artificial Intelligence, Machine Learning, Computer Vision, Cloud, Economics, Sustainability, AI, ML, Conversational AI, Natural Language Processing, NLP, Robotics, Security, Privacy, Information, Knowledge Management, Operations, Scientific Research, Search, Amazon, and Alexa
Updates
-
Announcing the Amazon Research Awards fall 2025 call for proposals. Successful applicants will receive unrestricted funds, AWS promotional credits, and training resources, including AWS tutorials and hands-on sessions with Amazon scientists and engineers. The deadline for submissions is November 5: https://amzn.to/4nXYcG2
-
-
Foundation models show promise for scientific computing, but adoption lags behind language and vision applications. Amazon researchers reveal what's needed: physical-constraint satisfaction, uncertainty quantification, and specialized forecasting techniques that maintain scientific rigor while overcoming data scarcity: https://amzn.to/488t0Q1
-
-
The Amazon Fulfillment Simulation Symposium brought together Amazon researchers from 61 teams across 7 countries to share breakthrough ideas in simulation-driven planning and fulfillment process optimization. The event showcased exciting work in artificial intelligence, digital twins, and innovative approaches to solving customer problems through simulation. Learn more about our work in operations research and optimization: https://amzn.to/46Cnl2d
-
-
Today, we're introducing the Nova Act extension—a tool that transforms how developers can build AI agents with Nova Act by bringing the entire agent development experience directly into IDEs like Visual Studio Code, Kiro, and Cursor. No more context switching between coding and testing—now you can generate scripts through chat, test cell-by-cell, and debug with live thinking and action logs all in one place. What used to take days now happens in minutes, helping developers build agents that achieve high reliability on enterprise workflows. As one partner at Hertz shared: "The live debugging alone cut our development time by 50%." Learn more: https://amzn.to/3KAM6nP
-
Agentic AI presents new scientific frontiers, like deciding what language agents should speak and modeling agentic negotiations. In this analysis, Penn professor and Amazon Scholar Michael Kearns explains how researchers will need to grapple with questions around sharing context without compromising privacy and understanding users' commonsense policies: https://amzn.to/4piTFzp
-
-
Audible's new ML-powered visual autocomplete system displays book covers as you type, reducing the number of steps to purchase. Learn how it's turning partial keystrokes into personalized book discoveries: https://amzn.to/4niePMp
-
-
Amazon's new Sensor Workbench simulator explores hundreds of warehouse sensor configurations in the time it takes to test just a few physical setups, using novel parallel-processing architecture and physics-based sensor modeling built on NVIDIA's Isaac Sim: https://amzn.to/47mPDzP