Writing
Notes from building real systems.
RAG vs OKF: two ways to give AI agents knowledge
How retrieval-augmented generation differs from Google's Open Knowledge Format, why agents are adopting curated knowledge, and when to combine both.Claude Code skills and subagents for real projects
How to teach Claude Code your project conventions with skills, delegate noisy work to subagents, and cut token usage on client projects.Building a production-grade RAG system: what matters
The layers that turn a RAG demo into a dependable product: ingestion, freshness, hybrid search, reranking, security, evaluation and cost control.Agentic AI frameworks: how to choose and best practices
A practical map of agent frameworks, from LangGraph to provider SDKs and low-code builders, and the engineering practices that make agents dependable.Why use AI? Practical value at work and in daily life
Where AI genuinely saves time at work and at home, where it fails, and the habits that make its answers useful and safe to rely on.Change data capture with Debezium and Kafka: SQL Server to MongoDB
How log-based CDC works, how Debezium turns SQL Server changes into Kafka events, and the patterns that keep a downstream MongoDB copy correct.Angular unit testing with Karma and Jasmine: a practical guide
Test Angular services, HTTP calls, standalone components and async code with Jasmine, Karma and TestBed, plus habits that keep a test suite fast.