FeedSources
⌘K
AS

Amazon Science Publications

100 articles total

Go to source

Amazon is a great place to practice science and have real business impact with publications. Amazon scientists continue to publish papers, teach, and engage with the worldwide research community.

Go to source
  •  FinLCR: Scaling financial long-context reasoning via curriculum reinforcement learning
  •  DODO: Discrete OCR diffusion models
  •  SEGRA: A structured experience guided reasoning agent for property graph question answering
  •  AgenTwin: An end-to-end framework for building agentic twins of mobile users
  •  SAS: Sparse attention synthesizer for efficient language model inference
  •  Conversation coach: A voice-enabled AI system that helps practice difficult workplace conversations
  •  Ishigaki-IDS-Bench: A benchmark for generating information delivery specification from BIM information requirements
  •  From demo to production: Generative search lessons from Alexa+
  •  PB-GRPO: Learning socially adaptive LLM agents from persona-driven simulation with preference-batched GRPO
  •  ControlsDSL: A language for verifiable cloud configuration controls
  •  Sequential multimodal evidence optimization for product media ranking in e-commerce
  •  IAM policy autopilot: Static analysis for policy generation from application code
  •  SDRF: A unified CDC abstraction for cloud-native SaaS data replication
  •  Attacking and defending multi-agent collaborative filtering systems through connectivity
  •  Fangorn: A conversational platform for geospatial intelligence analytics [demo]
  •  Non-parametric spatiotemporal trajectory prediction via state-conditioned transition sampling
  •  Testing GenAI models alongside model builds: A case study of collaboration between Amazon Nova and Chatterbox
  •  Pairwise ranking outperforms single-action RL for offline explanation selection: A practical lesson
  •  Multi-objective ranking for live-streaming: Balancing fresh and delayed signals with segment-aware targeting
  •  Analyzing seismic workflows with large language models: A case study on error detection