Amazon Science
@AmazonScience
Followers
93K
Following
11K
Media
663
Statuses
7K
The latest news and research from Amazon's science community. #AmazonScience
Global
Joined December 2019
🎥: Footage from this morning's LA-04 mission, as our latest batch of Amazon Leo satellites lift off aboard @ulalaunch's Atlas V.
4
34
202
Amazon's LLM-based system improves product listing quality by recognizing standard attribute values, collecting synonyms, and detecting errors. The process updates millions of listings within days:
amazon.science
Large language models are increasing the accuracy, reliability, and consistency of the product catalogue at scale.
0
4
14
Meet Amazon Nova 2 reasoning models Nova has introduced its next generation of foundation models that deliver frontier intelligence with industry-leading price performance 1/ Amazon Nova 2 Lite, is a fast, cost-effective model designed for everyday AI tasks. 2/ Amazon Nova 2
4
12
64
Amazon participated in the 2025 Winter Simulation Conference in Seattle last week. 700+ simulation leaders gathered for the 55th anniversary of the world's largest simulation event. Looking forward to continued collaboration on optimization and modeling.
0
2
7
Reminder: Trusted AI Symposium abstracts are due December 19. Papers using Amazon Nova models receive priority consideration. Submit today. #ResponsibleAI
amazon.science
Join leading researchers from academia and industry for an intensive, one-day symposium on the frontier of Trusted AI. This invitation-only gathering brings together distinguished scholars and...
0
2
8
Amazon's LLM-based system improves product listing quality by recognizing standard attribute values, collecting synonyms, and detecting errors. The process updates millions of listings within days:
amazon.science
Large language models are increasing the accuracy, reliability, and consistency of the product catalogue at scale.
0
4
14
🔊 New audio-processing technology from Amazon separates speech from background sounds using sub-band neural networks compressed to <1% original size. Results show 86% listener preference and 100% approval from users with hearing loss:
amazon.science
New audio-processing technology is making entertainment more accessible for millions of viewers.
1
1
10
AWS researchers present an approach to AI-native 6G: network language models creating intent-driven, verifiably safe, self-optimizing intelligence fabrics:
amazon.science
“Network language models” will coordinate complex interactions among intelligent components, computational infrastructure, access points, data centers, and more.
0
2
5
Since launching Amazon Nova Forge, we’ve had great conversations with the community about what it can do and why we built it. I explore the technical challenges we navigated to make "open training" a reality here: https://t.co/fe7NZT6VRR We built Forge initially because our own
0
3
18
Amazon Nova Forge enables organizations to build their own frontier models by mixing proprietary data with the data used to train Amazon Nova at each major stage of model development. More from @RohitPrasadAI on how the service works:
amazon.science
New service lets customers mix their own data with the data used to train Amazon Nova at each major stage of model development, enabling deep domain understanding while preventing "catastrophic...
0
2
14
NeurIPS ✅ We appreciated connecting with researchers across the science community. See you next year! #NeurIPS2025
0
1
20
Our researchers are presenting at several #NeurIPS2025 workshops today on topics spanning efficient reasoning, LLM security, and foundations of reasoning. Full schedule here:
amazon.science
The Neural Information Processing Systems (NeurIPS) annual meeting fosters the exchange of research on neural information processing systems in their biological, technological, mathematical, and...
0
1
6
At NeurIPS, researchers from @AmazonMusic are presenting work on how removing causal masking in transformer models improves music recommendations by 28% over deep learning baselines, effectively capturing both sequential patterns and stable user preferences. Learn more:
0
2
7
Looking to connect with Amazon researchers at NeurIPS? Several of our teams will be at our booth (#1523) today. Come say hi! 👋 9:30–10AM Alexa, Amazon Robotics, Fulfillment Technology and Robotics (FTR), Payment Risk ML 10–10:30AM AGI Autonomy, Alexa, AWS Field Experiences,
0
1
8
At NeurIPS, Amazon researchers are presenting MLZero, a multiagent architecture that separates data perception, tool knowledge, execution history, and code generation to enable machine learning automation that works with messy, real-world inputs. #NeurIPS2025
amazon.science
A multiagent architecture separates data perception, tool knowledge, execution history, and code generation, enabling ML automation that works with messy, real-world inputs.
0
1
8
The defending champions of the 2025 Amazon Nova AI Challenge are presenting their paper at NeurIPS. Check it out at today's poster session! #NeurIPS2025
Check out PurpCode at #NeurIPS next week, the first post-training recipe for safe code reasoning models! 🥇 Winner of 2025 Amazon Nova AI Challenge 🎯 Outperforms Sonnet 4, o4-mini, and R1 on cybersafety while maintaining utility Happy to chat!
0
0
8
At NeurIPS, Amazon researchers are presenting a neural certificate architecture that jointly represents inductive invariants and ranking functions for complete verification of reactive systems. Learn more about their work at today's poster session (#1409).
0
2
14
Don’t miss your chance to connect with us at NeurIPS. Stop by booth #1523 for live demos, technical talks, and conversations with our researchers. 📆 Full schedule: https://t.co/T42hh0Xu64
#NeurIPS2025
0
1
4
How do we get AI models to think more flexibly—like humans do?" Cognitive scientist @drperszyk is at NeurIPS this week talking about the frontier of adaptive intelligence and social world models. Learn more about her team's work at Amazon's AGI Lab: https://t.co/jshbsBL7tt
1
7
15