Articles

Technical articles, essays and historical notes, selected for English readers.

Search titles, full text and categories

All topics and categories

Archive layout

Page 3 / 4 · 46 articles

Technical guide

Published 15 June 2026 · 17 min

RAG, CAG, KAG, fine-tuning and instruction tuning: how do they differ, and which should you choose?

When a model gives an unsatisfactory answer, does it need more training, or simply access to the right information? Through practical analogies, this article compares RAG, CAG and KAG with fine-tuning and instruction tuning: retrieving documents, caching knowledge, reasoning over relationships and changing model behavior. A comparison table and real project scenarios help distinguish missing knowledge from unsuitable behavior before committing to training, and identify the method or combination that fits the task. Read article →

Security

Published 26 February 2026 · 8 min read

AI infrastructure security starts with the kernel and GPU

The AI stack connects the kernel, drivers, runtimes, containers, data, models and accelerators. A compatibility or trust failure in one layer can compromise the entire workload. Read article →

Security

Published 26 February 2026 · 6 min read

From Zero Trust Architecture to Zero Trust AI

Applying zero trust to the data, models and workflows of AI: why network access controls alone cannot protect the entire lifecycle. Read article →

Security

Published 24 February 2026 · 4 min read

MLOps as the foundation for Zero Trust AI

Continuous verification needs a traceable lifecycle for data, code, experiments, models and deployments. MLOps provides the operational foundation. Read article →

AI infrastructure

Published 19 May 2025 · 6 min

Choosing a GPU for AI: performance, cost and the actual workload

Compare useful performance against the full cost of deployment, including memory, software, the host server and the way the service will be used. Read article →

AI infrastructure

Published 19 May 2025 · 7 min

DGX, HGX or a PCIe GPU server?

Separate the need for a fast GPU fabric from the value of an integrated system, its software and its support contract. Read article →

AI infrastructure

Published 19 May 2025 · 6 min

Beyond the GPU: CPU, memory, storage and networking

Size the rest of a GPU server around data preparation, memory use, storage traffic and the application's communication pattern. Read article →

AI infrastructure

Published 19 May 2025 · 5 min

GPU types for AI: consumer, workstation and data center

How GPU product families differ in memory, software support and deployment requirements—and when each is a practical choice. Read article →

AI infrastructure

Published 19 May 2025 · 8 min

Choosing a PCIe GPU server: fit, topology, power and cooling

Check the exact GPU and server configuration, then verify sustained performance before accepting the system. Read article →

AI infrastructure

Published 19 May 2025 · 7 min

PCIe or SXM: choosing a GPU platform for AI

How memory, interconnects, power and expansion requirements affect the choice between PCIe cards and integrated SXM platforms. Read article →

Security

Published 13 May 2025 · 10 min read

A maturity model for Zero Trust AI

From manual model development to controlled automated processing: five stages for building the operational foundations of ZTAI and reducing access to sensitive data. Read article →

Security

Published 13 May 2025 · 8 min read

Zero Trust AI: principles and practical controls

Turn ZTAI principles into controls for identity, data, models, outputs and infrastructure, with explicit limits on what each control can establish. Read article →