Vitalii Ivanov
Senior Software Architect / Tech Lead
Here you'll find tutorials on technologies I recently found fascinating, or simply wanted to brush up on.
Tutorials
AI benchmarks: what the numbers actually measure
What one benchmark test looks like, who writes the tasks, who runs them, who grades the answers, and whether the model has already seen the questions.
The AI stack: how one claim crosses six layers
LangChain, LangGraph, Langfuse, Airflow and the Claude Agent SDK: which job each one owns, shown on one insurance claim.
Diffusion models: how noise becomes a picture
How a prompt and a grid of random numbers become an image, and why the network only ever predicts noise.
MCP: how one question becomes a tool call
How one question travels from your editor to a tool and back: the protocol every AI client now speaks.
Claude Code: how one prompt becomes a code edit
What a coding agent actually does between your prompt and the two lines it ends up changing.
Git worktrees: how three AI agents share one repo
How several agents work in the same repository at once without overwriting each other.
Interpreted vs compiled: how one line reaches the CPU
One line of code through Python, JavaScript, Java and Go, and why the split is a dial, not a dichotomy.
Reasoning models: how a model learns to think before it answers
What a model is really doing while it thinks, and why the thinking is what makes it right.
Concurrency: how Java, Node, Go & Python hold 10,000 clients
Four runtimes under the same flood of connections, and the two tricks all of them rely on.
Transformers: how five words become a sixth
How a handful of words becomes the next one: tokens, attention, layers, and a final vote.
LoRA: how to teach a frozen model in 8 megabytes
How to teach a frozen model a new skill by training a tiny patch instead of the weights.
Prompt injection: how a poisoned email hijacks an AI assistant
How an ordinary-looking email turns a helpful assistant against you, and the guardrails that stop it.
AI evals: how a one-line prompt change earns the right to ship
How a change to a prompt earns the right to ship, measured instead of guessed.
AI numbers every AI engineer should know
The handful of numbers that explain what a model costs, how fast it answers, and why.
AI agent patterns: how one support ticket gets closed
Every core agent pattern, shown on one real support ticket from open to closed.
How RAG works: retrieval-augmented generation
How a model answers from documents it never studied, and cites where the answer came from.
How a neural network reads a handwritten digit
Three kinds of network read the same hand-drawn digit, and each one sees it differently.
What happens when you type a URL and press Enter
Everything that happens between your keypress and the pixels: DNS, TCP, TLS, HTTP, rendering.