Next stop: Denver. Amazon will be at CVPR this week with workshops, live demos, and our latest computer vision research. Come find us at booth 349. #CVPR2026
Amazon Science
Research Services
Seattle, Washington 390,672 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
-
The first scalable flat-network datacenter is live. For decades, fat-tree topologies have dominated data center networking despite known inefficiencies: redundant router layers, top-of-tree congestion, and disproportionate failure impact. Random topologies were proven theoretically optimal in the early 1990s, but no one could make them work in practice. Routing demanded more memory than commodity hardware provides, and cabling a random topology across a physical data center was infeasible. RNG (resilient network graphs) solves both problems. It is now the default architecture for most new AWS data centers globally. The results: 69% fewer routers, up to 33% better throughput, and a projected 40% reduction in network equipment electricity consumption. The two key innovations, a passive optical device called a ShuffleBox and a routing protocol called Spraypoint, are detailed here: https://amzn.to/4dCkxqA
-
-
Meet the 68 researchers selected for the Amazon Research Awards fall 2025 cycle. Representing 49 universities across 11 countries, each recipient gains unrestricted funds, AWS Promotional Credits, and access to more than 800 Amazon public datasets and AWS AI/ML services. Explore the cohort: https://amzn.to/4wR1foK
-
Current LLMs overallocate to multilayer-perceptron layers. A new scaling law that factors in architectural choices identifies models with up to 47% higher inference throughput and no loss of accuracy. https://amzn.to/4fmkJv9
-
-
Building on a $110 million investment, the next generation of AI researchers are using Amazon's Trainium AI chip to accelerate discovery across ML, healthcare, and quantum computing – and all of it is open source: https://amzn.to/3PrhlVo
-
-
Reinforcement learning gyms, eval-driven model selection, and the shift from web-actuated to deterministic API-based agentic workflows are reshaping how AI shopping and event discovery systems reach production reliability. In this live SXSW panel, Amazon's Danielle Perszyk leads a discussion with leaders from Amazon and Bandsintown, examining how non-deterministic systems are moving from controlled environments into production. Watch on YouTube: https://lnkd.in/ejQqA58n