Digest of the Best Articles 2025: About Infrastructure, Development, and AI

Hello, tekkix! In this publication, we have gathered articles that our readers most often return to – on infrastructure, programming, AI, and management. These texts help to understand complex issues, fix production problems without panic, design systems more carefully, and view new tools with clear-headedness. If you are currently building infrastructure, optimizing databases and queries, implementing AI in work, or organizing processes and technical debt – proceed to the selection.

Infrastructure

Limits in NGINX: limit_conn, limit_req, worker_connections
This article explains how to limit load in NGINX and protect against overloads and DoS attacks. We will discuss what worker_connections sets, how limit_conn works for limiting connections, and limit_req for limiting request frequency. We will also look at config examples, dry-run tips, zone size, and response code 429.

K8s as a Foundation: Building an Infrastructure Platform
This article covers how to turn K8s into the foundation of an internal platform, rather than a collection of disparate clusters "on a whim." We will discuss the need for a platform team, how to deploy production, dev/staging, and CI/CD tools across different clusters, and how to automate the creation of new environments through IaC, pipelines, and cloud APIs.

Linux Administration Guide for Beginners
This is a guide for beginners on Linux administration: basic settings (hostname, timezone), commands for files and mounting, networking, user/group management, monitoring, and reading logs. At the end, readers will find file types and a mini cheat sheet.

To Keep the Network from Going Down... Again
An analysis of typical mistakes that cause the network to "drop": lack of planning and reserves for sockets/ports/addresses, overload of the "all-in-one" role on routers, unnecessary hops to servers, chaos with cables and labeling, plus lack of documentation and L1-L3 diagrams.

How to Design Resilient Event-Driven Systems at Scale
An article on how EDA really falls at peaks: back pressure, cold starts, throttling, and "retry boosters" are more important than mere latency. It provides resilience practices: queues/buffers + DLQ, shuffling sharding, dedicated resources for critical streams, fast-fail, backoff+jitter, event prioritization, and metrics/alerts.

Teaching Wazuh to Find Vulnerabilities
A practical analysis of the SCA module in Wazuh: how it checks host configurations against YAML policies (CIS), how rules and compliance work, where policies are stored on agents, how to enable/disable your files, and how to distribute them centrally from the server.

Postgres Migrations Using Logical Replication
A tutorial on migrating PostgreSQL with minimal downtime via logical replication: transferring the schema, setting up publisher/subscriber, publications and subscriptions, handling tables without PK (REPLICA IDENTITY), monitoring synchronization, and aligning sequences before switching.

Programming

Golang: When to Use Make, When to Use New
What is the difference between make and new in Go and why they are not interchangeable. It shows how slice/map/chan are structured within the runtime, why new gives a "zeroed out" instance while make gives a ready-to-use structure, and in which cases to choose each option.

Linus Torvalds: Critique of C++ — A Comprehensive Analysis
The article discusses Linus Torvalds' arguments against C++ in the Linux kernel: exceptions, RAII and "hidden" allocations, dependency on compilers and libraries, the risk of bloated abstractions. It also shows why predictability and control are more important for the kernel than the convenience of the language.

Bad JOINs: Techniques That (Accidentally) Take Down Production
An analysis of JOIN errors that take down production: Cartesian product, JOIN by functions, LEFT→INNER due to WHERE, different types, OR in ON, lack of indexes, fan-out. For each, how to rewrite the query, what indexes to add, and what to log.

Three mistakes in C# that are best avoided
An analysis of 3 typical bugs in C#: disabled nullable (NRE at runtime), hidden copies of struct in readonly context, and memory leaks due to unsubscribed events. Practical fixes are provided: enable Nullable + warnings as errors, mark struct/methods as readonly, unsubscribe/use weak events, and profile leaks.

Five PostgreSQL features that are rarely mentioned
PostgreSQL is a treasure trove of features that are often forgotten: we mark insert/update in upsert through RETURNING + xmax, grant rights on individual columns, get "the last row in the group" through DISTINCT ON, prohibit overlapping intervals with range + EXCLUDE, run one-time scripts through DO without functions.

Artificial Intelligence

The real use of generative AI in 2025
The article compiles fresh examples of how people are actually using generative AI in 2025 based on forums and publications. It has been found that therapy, life organization, and meaning-seeking are now in the lead, rather than just coding and tech support. Risks of dependency and privacy issues are also highlighted.

Computer vision for beginners
The article introduces CV in simple terms: the difference from image processing, basic terms, and stages of the pipeline (capture → preprocessing → features → recognition → interpretation). It briefly discusses tasks (classification, detection, segmentation) and the role of CNN, as well as provides recommendations to get started: Python, OpenCV, and NumPy.

AI is just a bag of words. How to stop seeing intelligence where it doesn't exist
We explore why we attribute "mind" and intentions to LLMs, and why this hinders a sober assessment of them. The author offers the metaphor of a "bag of words," which better explains hallucinations, strengths/weaknesses, and risks of such models.

Agents 101: How to create your first AI agent in 30 minutes
A step-by-step tutorial for beginners: how to build your first AI agent on LangGraph and CopilotKit in half an hour—from cloning the starter and API keys to running it locally, connecting to Copilot Cloud, and embedding chat UI in Next.js.

Management

Top 10 anti-patterns in software development to avoid
An article about development anti-patterns — from spaghetti code and God Object to copy-paste, "shotgun surgery," lava flow, dead code, "boat anchors," "magic buttons," and Big Ball of Mud. We'll discuss the symptoms that indicate their presence in the codebase, how they inflate technical debt and reduce maintainability, and what practices help eliminate them.

From Scrum Master to Delivery Manager: Evolution in the Age of Flow
This article explains why the role of Scrum Master logically evolves into Agile Delivery Manager: the focus shifts from "conducting Scrum rituals" to managing the flow of delivery and creating value. ADM coaches the team, removes blockers and dependencies, tracks flow metrics (lead/flow time, WIP, etc.), and helps build the value stream — more often in the logic of Kanban rather than "pure" Scrum.

6 Basic Concepts of Systems Thinking
An overview of six key concepts (interconnectedness, synthesis, emergence, feedback loops, causality, and system mapping) and why they are necessary to understand complex problems and find points of leverage.

Five Directions for Optimizing Business Processes
An article about 5 areas where efficiency often "leaks": manual input and paper, disconnected systems and data, poor communication between departments, heavy budgeting/forecasting/reporting, and slow period closing — with ideas on how to automate and speed this up.

Drawing Business Processes in Archimate Notation
How to describe business processes in Archimate at a high level: what 5 elements and connections are needed, how to denote roles, events, documents, and IT services, and why this simplified approach is more convenient than BPMN for architectural tasks.

More relevant knowledge from experts — at free demo lessons. Check out the event calendar and choose a topic for your task.