Blog
Notes on how I actually ship - stack choices, MVPs, and what a senior full-stack owns.
2026-10-21
Payment Reminder Bots for Utilities and Internet: Email, Telegram or Viber
A bot that reminds people about utility bills and internet payments in the channel they actually read - email, Telegram or Viber. Who orders it, how it works and what it takes to build.
Read →
2026-10-17
Which Online Games You Can Build as a Telegram Mini App
From one-tap arcade games to real-time multiplayer: which genres work inside Telegram, what the platform gives a game for free, and where the hard parts are.
Read →
2026-10-17
Your AI Chatbot, Fixed: Accurate Answers, Faster Replies, Lower API Bills
Why production chatbots give wrong answers, reply slowly and cost too much, and the five-step process I use to fix them: audit, retrieval, prompts, cost, report.
Read →
2026-10-14
Claude's Latest Models and How to Use an API Key Safely
A practical overview of the current Claude 5 family, which model to pick for which job, what changed in the API, and how to create, store and protect an Anthropic API key in a real project.
Read →
2026-10-12
Why I Prefer Long-Term Partnerships and Always Take a Project to Live
A launch is a milestone, not the finish line. Here is why long-term collaboration interests me and what I do to take every project all the way to production.
Read →
2026-10-08
Recovering a Lost Website Is Possible!
A missing website is not always gone for good. Learn how to check backups, hosting, domain access, and web archives and restore files and data safely.
Read →
2026-10-08
Developer Disappeared and the Project Is Unfinished? I Can Take Over Development and Support
An abandoned project does not always need a rewrite. I can assess its current state, secure access, prioritize the work, and take over development and ongoing support.
Read →
2026-10-07
Building an AI Assistant: From First Workflow to Production
A practical guide to scoping an AI assistant, preparing trusted knowledge, designing prompts and integrations, testing behavior, and operating it after launch.
Read →
2026-10-02
How to Evaluate an LLM Feature Before and After Launch
Build an evaluation loop around real user tasks, representative examples, measurable quality dimensions, human review, and production monitoring.
Read →
2026-10-02
Designing Long-Term Memory for LLM Agents
Separate conversation context from durable memory, then design extraction, retrieval, updates, forgetting, permissions, and evaluations as explicit system behaviors.
Read →
2026-10-02
How to Audit PaaS, IaaS, Networks, Firewalls, and Operating Systems
A practical, evidence-led method for reviewing cloud services, infrastructure, network paths, firewall rules, and operating-system hardening—and turning findings into a remediation plan.
Read →
2026-10-02
Eight Technology Topics Worth Writing About in 2026
A practical editorial shortlist for engineers: agentic development, AI security, software supply chains, observability, multimodal systems, and more.
Read →
2026-09-28
Building with Supabase: The Engineering Details That Matter
Supabase speeds up product development, but production quality still depends on deliberate database design, Row Level Security, server-side boundaries, migrations, and operational discipline.
Read →
2026-09-23
How I Build Digital Products for the European Union: Data, Rights, and AI
A personal engineering standard for EU-facing projects: respect the rights attached to digital materials, minimise personal data, and assess AI systems against the rules that apply to their purpose and risk.
Read →
2026-09-23
Leading AI models in 2026: price, quality and the cost of a useful answer
GPT-6, Claude, Gemini and DeepSeek compared: API prices, practical trade-offs and a common cost calculation for ten models. Updated September 23, 2026.
Read →
2026-09-21
Payment Platform Security: Critical Protections for Fiat and Cryptocurrency Systems
Whether processing cards through Stripe or accepting Web3 crypto transactions, payment security is the ultimate test of product engineering. Learn why it is non-negotiable, how to design protection for fiat and crypto, and how to safeguard user funds.
Read →
2026-09-21
Why Auditing an Inherited Project Before Work Is Critical: Saving Budget, Timelines, and Sanity
Taking over a project without a thorough technical audit means inheriting hidden risks. Why Product Engineers should audit before work begins, and how to do it.
Read →
2026-09-21
The Product Engineer Mindset: Shipping Value Over Code Volume
Why leading founders and scale-ups hire Product Engineers who own outcomes from the customer problem to database queries and product analytics.
Read →
2026-09-20
Shipping AI-Native Products as a Product Engineer: Practical LLM Architecture
How Product Engineers turn unpredictable LLM capabilities into dependable production software through structured JSON outputs, streaming UX, and deterministic fallback circuits.
Read →
2026-09-20
Technical Debt & Velocity: How Product Engineers Balance Speed and Scalability
How to make intentional technical debt decisions that accelerate time-to-market without compromising long-term maintainability and system stability.
Read →
2026-09-19
Model Context Protocol (MCP) & Multi-Agent Workflows in Production
Why Model Context Protocol (MCP) is becoming the USB-C standard for LLMs, how multi-agent graphs replace monolith prompts, and how to build reliable production agent loops.
Read →
2026-09-19
RAG 2.0 & GraphRAG: Building Enterprise Knowledge Engines That Don't Hallucinate
Naive vector search is dead. Discover how hybrid retrieval, Cohere reranking, GraphRAG, and multimodal parsing eliminate hallucinations in production AI applications.
Read →
2026-09-19
Local LLMs, vLLM, and LoRA Fine-Tuning: Escaping Vendor API Lock-in
How high-throughput inference engines like vLLM, quantization, and LoRA fine-tuning allow engineering teams to run private 8B-70B models with enterprise-grade performance.
Read →
2026-09-19
Real-Time Voice AI and Multimodal Pipelines: Building Sub-300ms Interactive Apps
Architecting low-latency voice and vision interfaces using WebRTC, OpenAI Realtime API, Gemini Live, and event-driven Node.js microservices.
Read →
2026-09-19
LLMOps, Security Guardrails, and Cost Routing: Enterprise AI Infrastructure
How to prevent prompt injection, enforce data privacy (PII scrubbing), control API spending with semantic routers, and maintain robust LLMOps pipelines.
Read →
2026-09-19
Harness.io: Complete Overview of the AI-Powered Software Delivery Platform
From scriptless CI/CD and AI-driven automated rollbacks to Cloud Cost Management (FinOps), Security Testing Orchestration (STO), and Internal Developer Portals (IDP). A detailed breakdown of Harness.io features, Delegate architecture, and enterprise benefits.
Read →
2026-09-18
Quality Gates: testing and engineering contracts in the age of AI agents
When AI generates code 10x faster than humans can read it, manual code review breaks down. How Quality Gates and automated testing turn AI agents into reliable teammates — insights from Ilya Klimov's engineering philosophy.
Read →
2026-09-18
Agentic Coding and reality: why prompting is not enough for real software
Moving from raw prompt engineering to structured Agentic Workflows, MCP servers, and bounded graphs. Reflections on Ilya Klimov's critique of the 'Dark Software Factory'.
Read →
2026-09-18
Architecture in the era of AI: quotes and takeaways from Ilya Klimov's Fwdays talk
Is software architecture dead in the age of vibe coding? Key quotes from Ilya Klimov's talk at JavaScript fwdays'25 and my take as a Senior Full-Stack & AI Engineer.
Read →
2026-09-10
Loop agents vs graph agents: how I choose the architecture
Every agent is a model calling tools in some order. The real question is who decides the order - with diagrams and a side-by-side comparison.
Read →
2026-09-10
Why I love AI - and everything around it
Not the hype. The new kinds of products, the leverage, the engineering puzzle - and the feeling of being a beginner again every few months.
Read →
2026-09-10
LangChain: interesting facts and when I actually reach for it
A side project older than ChatGPT, a $1.25B company, a 1.0 rewrite - and an honest take on when a framework helps and when it just adds layers.
Read →
2026-09-10
Llama: the model that leaked, started the open-weights movement, and then changed course
From a 4chan torrent to a billion downloads and a closed successor - the most interesting facts about Meta’s Llama, and what they mean for builders.
Read →
2026-09-10
Who is a product engineer - and why I call myself one
Not a résumé title. An engineer judged by what users can do, not by how many tickets were closed.
Read →
2026-09-10
TypeScript + Python: one engineer, two ecosystems, one contract
TypeScript for the product, Python for the models - and the habits that keep the two from drifting apart.
Read →
2026-09-10
The values of a TypeScript developer
Types are not decoration. They are the cheapest documentation, the first test, and the contract your team reads at 2 a.m.
Read →
2026-09-10
How I use Next.js at work
Server first, client where it pays, and a cache I can explain. The defaults I bring to every Next.js product.
Read →
2026-09-10
Building AI products and real-estate platforms: same rules, different data
Listings, leads, viewings, and a model that helps an agent instead of pretending to be one.
Read →
2026-09-10
Kubernetes in my work: when it earns its place and how I use it
Kubernetes is not a badge. It is what I reach for when one server and a process manager stop being enough.
Read →
2026-09-10
How games showed me I love programming - and why I love AI
A game is a system with rules, feedback, and a loop you want to replay. So is good software. AI just made the loop faster.
Read →
2026-09-10
How AI pushed a TypeScript engineer into Python
Node.js is where my products live. Python is where the models live. Learning it was not a career switch - it was following the work.
Read →
2026-09-10
How I build Telegram Mini Apps
A Mini App is a real web product that happens to open inside a chat. Auth, theme, payments, and the parts Telegram gives you for free.
Read →
2026-09-10
React 19.3 vs React 18: what actually changed for a product team
View Transitions, Fragment refs, browser(), Activity, the Compiler - and which old workarounds you can finally delete.
Read →
2026-09-10
SQL is the contract that outlives your ORM
Frameworks change every few years. The schema, the constraints, and the query plan are what still hold when real money hits the table.
Read →
2026-09-06
How I wire an LLM into Node.js - and what I refuse to ship
The model is a remote call with a bill. The product is still your API, your queue, and one action a human can trust.
Read →
2026-08-12
How I scope an MVP - and what I refuse to add
A first version is one action in production, not a platform. The five filters I use before I open the repo.
Read →
2026-07-03
What a senior frontend actually owns in a product team
Not the pixel. The path: data, states, tests, and the night someone will page you.
Read →