Deepgram enhances Amazon SageMaker AI support with AWS IAM Temporary Delegation
Amazon/AWS
Deepgram has integrated AWS IAM Temporary Delegation into its support workflow for self-hosted speech AI models on Amazon SageMaker AI. This allows customers to grant scoped, time-limited access to Deepgram engineers for troubleshooting, reducing initial investigation time from days to minutes while maintaining auditability.
Deepgram announced integration of AWS IAM Temporary Delegation to improve support for enterprises running self-hosted speech AI models on Amazon SageMaker AI. Previously, troubleshooting required scheduling screen-shares or provisioning long-lived cross-account IAM roles, which was slow and auditable. With the new integration, a Deepgram engineer can request access via a support ticket, the customer approves in their IAM console, and temporary STS credentials are issued for a specific SageMaker AI endpoint, log group, and region, expiring after 12 hours. The customer retains full control and can revoke access anytime. All delegated API calls are logged in the customer's AWS CloudTrail. Deepgram supports models Nova, Flux, and Aura-2 on SageMaker AI via AWS Marketplace.
- Сокращения
- IAM = Identity and Access Management — Управление идентификацией и доступом
- ARN = Amazon Resource Name — Имя ресурса Amazon
- STS = Security Token Service — Служба токенов безопасности
- SNS = Simple Notification Service — Простая служба уведомлений
- GPU = Graphics Processing Unit — Графический процессор
- STT = Speech-to-Text — Распознавание речи в текст
- TTS = Text-to-Speech — Синтез речи из текста
Source: AWS ML blog —
original