← All organizations

Retail, cloud computing and AI services

Amazon

Amazon sells consumer goods and provides cloud computing and AI services through Amazon Web Services (AWS). Its Amazon Bedrock service gives businesses access to managed foundation models from multiple providers, with governance and security controls; Bedrock AgentCore supplies infrastructure for deploying and operating agents. For consumers, Alexa for Shopping offers product comparisons, personalized recommendations, price history and automated purchasing features.

Jeff Bezos founded Amazon in 1994, initially as an online bookstore. Andrew R. Jassy, previously AWS CEO, has led Amazon as president and CEO since 2021. Its engineering contributions include the 2007 Dynamo paper, which described a highly available key-value storage system used by core Amazon services. Dynamo uses object versioning and application-assisted conflict resolution, accepting weaker consistency under certain failures to maintain availability—a concrete example of designing distributed systems around operational tradeoffs.

Amazon’s business infrastructure extends beyond computing: Amazon Supply Chain Services lets other businesses move, store and deliver raw materials and finished products using the supply chain supporting Amazon itself. Early customers include Procter & Gamble and 3M. In July 2026, the company reported hundreds of thousands of Amazon Bedrock customers, illustrating the reach of its managed model platform alongside its retail and logistics operations.

www.aboutamazon.com

Start here

  1. Building Blocks for LLM Systems & Products

    Start here to understand how evaluation, retrieval, safety checks, and user feedback fit together in a production LLM product.

    Eugene YanAI Engineer Summit 2023

  2. Recsys Keynote: Improving Recommendation Systems & Search in the Age of LLMs

    Learn how language-model techniques can change item representation and training data for recommendation and discovery systems.

    Eugene YanAI Engineer World's Fair 2025

  3. Agent Output Is Not UX: Rendering Layer Your LLM Pipeline Is Missing - Bala Ramdoss, Amazon Lens

    Understand the tradeoff between controlled components and open-ended generated interfaces, and where a Backend-for-Frontend service handles safe fallbacks.

    Bala RamdossAI Engineer World's Fair 2026

  4. Spec-Driven Development: Agentic Coding at FAANG Scale and Quality — Al Harris, Amazon Kiro

    Amazon principal engineer Al Harris demonstrates how Kiro uses spec-driven development to turn prompts into structured requirements, acceptance criteria, design artifacts, and implementation tasks.

    Al HarrisAI Engineer Code 2025

Messages from the stage

Evaluating products beyond the demo

Eugene Yan's building-blocks talk connects task-specific evaluations, retrieval, guardrails, and feedback-driven design. A joint discussion featuring Yan extends that foundation to monitoring, provider changes, and team workflows.

Language models for recommendation and search

Eugene Yan examines semantic item IDs, generated metadata and training data, and models that span multiple tasks. The examples connect multimodal item representations with cold-start problems and search quality.

Adding structure around agent output

Al Harris demonstrates explicit requirements, acceptance criteria, and design artifacts for agentic coding. Bala Ramdoss addresses the delivery side through version-aware rendering contracts, structured-UI streaming, and platform-specific rendering.

Affiliations reflect each recorded session, not necessarily current employment.

Company sources · checked 2026-08-28