LitMy.ru - литература в один клик

Scalable AI Systems (Early Release)

  • Добавил: literator
  • Дата: 8-07-2026, 21:13
  • Комментариев: 0

Название: Scalable AI Systems: Architecting from Prototype to Production-Ready AI (Early Release)
Автор: Swagata Ashwani, Ashwani Nandini
Издательство: O’Reilly Media, Inc.
Год: 2026-07-08
Язык: английский
Формат: pdf, epub
Размер: 10.1 MB

AI and Generative AI systems in production present challenges that traditional software practices weren't built to handle. They behave probabilistically, evolve over time, and introduce risks around quality, bias, security, and regulatory compliance that require a fundamentally different approach. To build systems that perform reliably at scale and earn trust from users, regulators, and organizations, teams need to embed governance into architectural decisions from the start, not treat it as an afterthought.

In Scalable AI Systems, Swagata Ashwani and Ashwani Nandini give you a unified framework for connecting system architecture with evaluation, assurance, governance, and compliance, guiding you from prototype to production-ready AI. Reference architectures, checklists, and case studies drawn from regulated, legacy, and digital-native environments provide ready-to-use resources you can apply at every stage of the AI lifecycle, whether you're designing a new system or improving one already in production.

Design scalable, reliable, production-ready AI and GenAI systems
Define quality gates, evaluation strategies, and monitoring metrics
Connect architecture decisions to fairness, explainability, risk, and compliance
Prepare your systems and organization for audits, regulatory requirements, and long-term governance
Align engineering, QA, risk, and product teams around shared terminology, responsibilities, and quality standards

Скачать Scalable AI Systems (Early Release)












[related-news] [/related-news]
Внимание
Уважаемый посетитель, Вы зашли на сайт как незарегистрированный пользователь.
Мы рекомендуем Вам зарегистрироваться либо войти на сайт под своим именем.