The parts nobody explains.
AI integration, frontend architecture, and the dev tooling decisions that actually matter at scale — with code, war stories, and no filler.
Start Here
Reader picks worth starting with
What is MCP? The Protocol That Finally Connects AI to the Real World
MCP (Model Context Protocol) is the USB-C of AI tooling — a universal standard that lets any AI model talk to any data source or tool without custom glue code per integration. If you've ever wondered how Claude 'knows' what's in your files or can run a terminal command, this is the protocol doing that work.
Building Your First MCP Server: Tools, Resources, and the Right Mental Model
Building an MCP server is simpler than it looks — a few tool definitions, a request handler, and a stdio transport. The hard part is designing tools the model will actually use correctly. This guide builds a real server from scratch and covers every design decision that separates a good server from a frustrating one.
The Mental Model That Finally Made Angular Signals Click
If Angular Signals still feel 'almost right but oddly off', this graph-first mental model explains why Hooks analogies fail and how dependency tracking actually works.
Articles
Deep dives and practical guides
What is MCP? The Protocol That Finally Connects AI to the Real World
MCP (Model Context Protocol) is the USB-C of AI tooling — a universal standard that lets any AI model talk to any data source or tool without custom glue code per integration. If you've ever wondered how Claude 'knows' what's in your files or can run a terminal command, this is the protocol doing that work.
Building Your First MCP Server: Tools, Resources, and the Right Mental Model
Building an MCP server is simpler than it looks — a few tool definitions, a request handler, and a stdio transport. The hard part is designing tools the model will actually use correctly. This guide builds a real server from scratch and covers every design decision that separates a good server from a frustrating one.
I Used AI for Angular Architecture Reviews for 3 Months. Here's What Changed.
Real-world lessons from using AI as an architectural reviewer: what improved, where AI was weak, and how to structure prompts for useful feedback.
You're Using Claude Wrong as a Developer
10 power moves and 5 bonus hacks that changed how I ship code. From treating Claude like a fancier Stack Overflow to unlocking its full potential — concrete prompts, real examples, no fluff.
The Frontend Engineer's Honest Guide to Gen AI
From skeptic to daily user — an honest take on how Gen AI actually shows up in frontend work without the hype. What LLMs really are, what they're good at, and what frontend devs still need to own.
AI as a Search Engine vs a Thinking Partner: The Difference That Changes Everything
Why prompt quality and interaction style matter: shifting from answer retrieval to collaborative reasoning changes both speed and depth of engineering work.
Claude Code: Using AI to Build and Manage Your Design Token System
Design tokens are tedious to name, hard to keep consistent, and painful to scale across themes. Here's how to use Claude to generate a full two-tier token system, map dark mode, audit naming drift, and convert between formats — with practical prompts you can use today.
Opencode Tokens: How They Work and How to Stop Burning Them
Every turn in opencode sends the full context window — history, file reads, tool outputs, all of it. Here's exactly how tokens accumulate and the concrete steps to cut your burn rate without losing productivity.
Vibe Coding: AI-First Development Is Reshaping Frontend Engineering
You describe a feature in plain English and get working JSX, typed services, and passing tests. Vibe coding is not a gimmick — here's the workflow, the tools, and what frontend devs must still own.
Streaming UX: Building Real-Time AI Interfaces That Feel Alive
A spinner for 8 seconds is a dead UI. The same wait with tokens streaming in feels fast. SSE, Angular signals, the cursor pattern, and every UX detail that separates polished AI products from janky ones.
LLM API Cost Guide: What You Actually Pay in 2026
Real pricing across Claude, GPT-4o, and Gemini — cost scenarios for chatbots, code review tools, and side projects, prompt caching savings, model routing strategy, and how to instrument costs before your bill surprises you.
On-Device AI: Running LLMs Directly in the Browser
Zero API cost, offline-capable, and data never leaves the device. WebLLM, WebGPU, Chrome's built-in AI APIs, and the hybrid architecture pattern for production apps.
Angular + MCP: Building a Model-Aware Frontend
How to wire Angular signals, streaming responses, and MCP tool use into a production-grade AI code review assistant — typed end to end.
Building AI Agents with Tool Use: From Concept to Production
The ReAct pattern, agentic loops, tool design principles, multi-agent orchestration, and the safety guardrails you need before shipping.
RAG Architecture Deep Dive: Building Context-Aware AI Applications
Chunking strategies, hybrid search, reranking, HyDE query transformation, and the RAGAS evaluation framework — the complete RAG engineering guide.
Prompt Engineering for Developers: From Basics to Production
System prompt architecture, few-shot patterns, chain-of-thought, structured output, prompt injection defense, and building an eval loop.
AI Terminology for Developers: A Practical Glossary
Tokens, embeddings, context windows, RLHF, RAG, temperature, chain-of-thought, tool use, MCP — every term you actually need, explained for engineers.
AI-Powered Developer Tools: Reshaping the Engineering Workflow
From code completion to autonomous agents — how AI is becoming an integral layer of the modern development stack.
Learning Roadmap
Navigate from beginner to advanced
AI & LLMs
From AI concepts to production agents
Frontend Engineering
Modern web architecture patterns
Dev Tools & Workflow
Maximize engineering velocity
Newsletter
Stay in the loop
New writing on frontend engineering, system architecture & AI — delivered straight to your inbox. No spam, unsubscribe anytime.