Skip to content
alex.fialko
RU
Prague · Full-Stack & AI EngineerOpen to projects

AI systems, bots and products that actually reach production

I'm Alex Fialko. I build AI agents, Telegram bots and full-stack products — from architecture to deploy, on my own. Not pilots, not demos for the sake of demos.

Message on TelegramSee the work
19
projects with the decisions explained
5
with a live version
8
in private repos — architecture shown

Work

What I've built

Not a list of technologies: for each project — the problem, the solution, and what happened in practice, including what broke.

Featured

AI systems

Debrief

Meeting minutes from two tracks instead of diarization

15 мс
track drift over 12 s
9.9×
faster than realtime
35
tests
Python 3.11+PyAudioWPatch (WASAPI loopback)faster-whisperAnthropic structured outputs
AI systems

Pulse

A watchdog over a dozen services that doesn't become noise

6
real addresses exercised
40
tests
3.11–3.13
Python in CI
Python 3.11+ (asyncio)aiogram 3httpxRailway GraphQL API
AI systems

LeadRadar

A Threads lead pipeline that survives a crash

2
BullMQ queues
24 ч
score cache TTL
TypeScript (strict, ESM)BullMQ + RedisPlaywrightGoogle Gemini
AI systems

Agent Farm

Hub-and-spoke agent farm with isolated RAG workspaces

5
rollout stages
0
shell tools given to the model
TypeScriptNode.js 20FastifyBullMQ

More projects

What I do

Four areas, one approach

01

AI agents & automation

LLM integrations, multi-agent pipelines, monitoring and data processing with no human in the loop where that's justified.

02

Telegram bots & Mini Apps

From a simple command bot to one with AI logic, a database and cron jobs in production.

03

Full-stack web products

Next.js/React on the front, Node or Python on the back, honest work with data and deployment.

04

Turnkey product MVPs

From idea to a working version in the hands of first users — fast, without extra layers of abstraction.

How I work

No surprises in the process

  1. 01

    Discovery

    We work out what actually needs solving — before the code editor opens.

  2. 02

    Architecture

    I pick the stack and architecture for the task, not out of habit. Decisions and their reasons are written down.

  3. 03

    Build

    Iteratively, with working intermediate versions — progress is visible every week, not at the end.

  4. 04

    Deploy & iterate

    Production isn't the finish line, it's the start. I monitor, fix and refine based on real use.

About

An engineer, not just a “vibe coder”

Based in Prague, building systems where LLMs meet ordinary engineering. Every project on this page was built by me alone: architecture, code, tests, deployment.

I use AI tools not because it's fashionable but because they really shorten the path from idea to working code — provided you understand what they do and where they can lie.

Principles you can see in the cases

  1. 01

    A deterministic core around the model

    The dice roll, the alerting, the limits of authority live in code you can test. The model explains; it doesn't decide.

  2. 02

    Checked on real hardware

    Tests didn't reveal that WASAPI goes silent without sound. A live run did. I look for those places before the client does.

  3. 03

    Hard limits on autonomy

    No auto-publishing, no outbound messages behind a human's back, no shell for the model — as an invariant, not a setting.

Open to partnerships and projects that need someone who will carry the architecture through to the end, not drop it halfway.

Contact

Tell me what needs building

Projects, partnerships, consulting — write directly, no forms, no approvals. I reply fast.