Cloud & Infrastructure
How NVIDIA OpenShell Reference Architecture Secures and Governs AI Agent Workloads at Scale
NVIDIA's OpenShell Reference Architecture is designed to secure and govern AI agent workloads at scale. These autonomous agents can perform complex tasks, access files, tools, the internet, and even spawn other agents, which increases risks such as downloading code and requesting additional privileges. The architecture helps manage these risks.
- Advanced (300)
Build a semantic layer for agentic AI on AWS with Denodo and Amazon Bedrock AgentCore
AWS, in collaboration with Denodo and Amazon Bedrock AgentCore, has developed a governed semantic layer for agentic AI. This layer federates data from Amazon Aurora, Amazon Redshift, and an on-premises PostgreSQL database into a unified customer 360 view, enabling queries via VQL with identity propagation and access control.
- Amazon Bedrock
How Octave built a secure multi-tenant AI foundation on Amazon Bedrock
Octave has built a secure, multi-tenant AI foundation on Amazon Bedrock, utilizing Cohere Embed v4 for embeddings and Amazon Bedrock AgentCore for isolated agent execution. This setup supports a Complaint Triage Assistant that searches 2 billion regulated documents without crossing tenant boundaries.
- Advanced (300)
How Eon built multi-cloud investigation with an MCP Gateway on Amazon Bedrock AgentCore
Eon has implemented a unified Amazon Bedrock AgentCore gateway to replace per-tool credentials and local MCP processes across AWS, GCP, and Azure. This has streamlined production investigation access for over 100 developers and reduced new engineer onboarding time from two days to two minutes.
- Advanced (300)
Securing AI agent identity with X.509 PKI on AWS
The article discusses securing AI agent identity using X.509 PKI on AWS, leveraging Keyfactor EJBCA and AWS services. It details a reference architecture that uses signed request envelopes, user-signed delegation assertions, RFC 8693 audience-scoped tokens, and real-time OCSP revocation to ensure agent identity and prevent request forgery.
- Advanced (300)
Fine-tuning GPT-OSS-20B for multilingual reasoning on Amazon SageMaker HyperPod
The AWS blog post discusses fine-tuning the GPT-OSS-20B model for multilingual reasoning using Amazon SageMaker HyperPod. It highlights the use of supervised fine-tuning (SFT) and reinforcement learning with GRPO, with SFT improving reasoning-language accuracy from 22% to 100%.
- Advanced (300)
Beyond the KV cache: Training and serving i14’s Laplace 3B linear attention model on AWS Trainium2 with native PyTorch
i14 trained and served the Laplace 3B linear attention model on AWS Trainium2 using native PyTorch. The model maintains a fixed recurrent state of 29.5 MB per sequence, avoiding the growing key-value cache of transformer models and ensuring predictable memory usage.