Axel Antonín Málek
EN CS

Antonín Málek · Axel

I like taking things apart to see how they work.

I'm Antonín Málek — I go by Axel internationally. I write software and spend a lot of time trying to understand the systems underneath it.

Currently building a self-hosted transit app, a desktop assistant, and smaller tools around maths, security and AI. Learning in public, one project at a time.

See my projects LinkedIn GitHub

About

Curiosity first, then the code

I learn best by building. When something catches my interest I usually try to make a small version of it myself — a transit app, a scanner, a game, a neural network written from scratch — because the parts I don't understand become obvious the moment they have to work.

That is why my projects are spread across quite different areas. A transit backend forces me to think about real, messy data and honest failure states. A security tool makes me read systems the way an attacker would. Writing a network by hand, rather than importing one, is how the mathematics finally stops being a black box.

I am not claiming to be an expert in all of this. Some of it is years of practice, some of it is a few weeks of curiosity, and some of it is on a list I haven't started yet. I try to label those honestly, because knowing the difference is part of learning.

What ties it together is the same instinct each time: I want to know how a thing works underneath, and I would rather understand one difficult idea properly than collect a hundred easy ones.

At a glance

  • Goes by Axel
  • Comfortable with Python, C#, Linux
  • Actively learning C, frontend, security
  • Happiest when Debugging something real

Projects

Things I've built

A mix on purpose: a few larger ongoing builds, some finished small tools, and honest experiments. Each card says what it is, what it is made of, and how finished it actually is.

Showing 14 of 14 projects

LIVE VEHICLES
01 Active now

Bussa / Swift — self-hosted transit backend

A self-hosted backend and web app for public transport in Pardubice and Hradec Králové. One process serves a JSON API, live vehicle positions, stop arrivals, multi-leg trip planning and an admin panel.

The interesting part Several unrelated city data sources are normalised behind one API, with deterministic C# logic doing all routing and timing. A local LLM only rephrases answers from structured data — it is never allowed to invent a departure time.

  • C#
  • .NET 10
  • ASP.NET Core
  • React
  • Vite
  • MapLibre
  • SQLite
Source
OSONLINESPOTIFY
02 In progress

Jarvis AI

A Python desktop assistant inspired by the films: it can talk to online services, control parts of the operating system, and drive a Spotify player through separate function modules.

The interesting part The hard part is not the voice, it is the plumbing: cleanly separating 'online' actions from 'operating-system' actions behind small, testable function modules so the assistant can grow without turning into one giant script.

  • Python

Long-term, ambitious goal — far from finished, and I won't pretend otherwise.

Source
v = 7.66 km/s● TARGET LOCKEDFRAME 0341 · CONF 0.9430 FPSΔp 12.4 px
03 Experiment

ISS speed tracking

A small Flask site that measures and graphs the speed of the International Space Station from imagery, with a second view that plots the result over time.

The interesting part Turning pixels into a physical quantity: the interesting problem is computing velocity from tracking data and making the reading stable rather than jumpy between frames.

  • Python
  • Flask
  • OpenCV
  • HTML
  • CSS

The public GitHub copy is a work-in-progress, not the finished tool — the complete version is kept private. The visual below is an illustration of the tracking approach, not a live feed.

Source
04 In progress

Security & network tooling

Hands-on security work: a packet/IPX tool written in C#, plus practical network scanning with Nmap and CVE lookups to understand how services and misconfigurations actually show up on a wire.

The interesting part Reading scan output critically is the real skill: distinguishing an outdated banner from a genuine vulnerability, and mapping findings to CVE identifiers instead of guessing.

  • C#
  • Python
  • Nmap
  • NVD / CVE

The preview below is a self-contained illustration with fake hosts and made-up CVE numbers — it does not scan anything and exposes no real data.

Source
.DV ORIGINAL AVI (lossless)MP4 (H.264)original .dv FireWire / i.LINK · capture → keep → export
05 In progress

Digital8 Studio

A cross-platform desktop application for capturing video from a Sony Digital8 camcorder over FireWire, always preserving the untouched original and exporting AVI and MP4 copies.

The interesting part Hardware I don't have is the interesting constraint: the app uses an honest three-state capability model and refuses to guess, reporting 'unknown' rather than fabricating a result for a device it has not seen.

  • C#
  • .NET
  • Avalonia
  • FFmpeg
  • FireWire

Local project — no public repository yet.

HELLO TIMING MODEL
06 Experiment

Typing assistant overlay

A small overlay tool that studies how text is typed and replays it, with a separate engine for typing content and a model of realistic timing.

The interesting part Human-like timing: instead of a constant delay, the timing model varies key presses so the output reads like real typing, which is a neat little modelling problem.

  • C#
  • .NET
  • Avalonia

Local project — no public repository yet.

hi there wayland overlay · window edges = ground
07 Experiment

stickmen

Little stick figures that live on top of a Wayland desktop. They treat window edges as ground, have individual personalities and memories, chat with each other, and occasionally brawl.

The interesting part Perception and simulation in one loop: the figures poll real window geometry, react to the cursor, and read the screen through a local OCR and vision pass — all of it best-effort and honest about its own uncertainty.

  • Python
  • GTK
  • Cairo
  • Wayland
  • Ollama

Local project — no public repository yet.

FORWARD PASS
08 Learning project

Neural network experiments

Small from-scratch neural networks written in C# to understand the maths directly, including a hand-built perceptron trained on a simple logic problem.

The interesting part No frameworks: implementing weights, forward passes and learning by hand makes the idea of 'training' concrete instead of a black box.

  • C#
Source
09 Experiment

Life Game — artificial life in C#

A cellular-automaton experiment in C# exploring how simple local rules grow into surprisingly complex, lifelike behaviour.

The interesting part Emergence: the rules fit on a few lines, yet the patterns that appear are rich enough that watching them evolve is the whole point.

  • C#
Source
∫ f(x) dx
10 Finished

Math & physics calculators

A set of small tools built while learning: an integral calculator, a projectile-motion calculator, a visualisation of pi, and a peptide/chemistry calculator.

The interesting part Turning textbook formulas into working code — a good way to find the gaps in my understanding, because the program refuses to be vague about the maths.

  • C#
  • Python
Source
CANVAS · 60 FPS
11 Finished

Early web experiments

Where the frontend interest started: a hand-written first website, a Canvas flying-bird game, a weather-forecast script and a webcam page that taught me how pages, styles and requests fit together.

The interesting part Learning the fundamentals properly: the goal was to write it myself rather than reach for a framework, so the lessons would stick.

  • HTML
  • CSS
  • JavaScript
  • Canvas
  • Python
  • Flask
Source
12 Finished

Small RPG project

A small role-playing game written in C#, later rebuilt with a graphical version — one of my larger early programming projects.

The interesting part Structure for the first time: keeping state, rules and content apart in a program that grew beyond a single screen of code.

  • C#
Source
WHERE IT STARTED
13 Finished

drak pavel — my first bigger project

The project I keep as a marker of where all of this started, written at the very beginning of my programming. It is rough, and that is exactly why I keep it.

The interesting part Finishing something at all: the real lesson was that a complete small program teaches more than an ambitious unfinished one.

  • C#
Source
14 Private

Driving-school website (private)

A website I am currently building for a driving school. It is private work, so no preview, screenshots or code are shown here.

The interesting part Building for a real client raises the bar: content, clarity and reliability matter more than novelty.

  • HTML
  • CSS
  • JavaScript
  • C#
No public preview — private work

Exploration

Areas I keep returning to

These are not separate hobbies, they overlap. Pick a node to see how it connects to the rest.

CURIOSITY Software Security AI Science Creative

Select a node — then use the arrow keys to move between them.

I use this regularly I'm actively exploring

Designing programs and understanding what runs beneath them.

  • C# / .NET
  • Python
  • APIs & JSON
  • Linux internals
  • C

Reading systems the way an attacker would, and giving back to the tools I use.

  • Linux daily driver
  • Open-source tooling
  • Network scanning
  • Application security

From hand-written perceptrons to local language models.

  • Neural nets from scratch
  • Local LLMs
  • AI-assisted development
  • Computer vision

The ideas I read about for the pleasure of understanding them.

  • Calculus & integrals
  • Physics
  • Quantum mechanics
  • Scientific computing

The slower things that keep the fast ones in balance.

  • Music
  • Philosophy & reading
  • Chess
  • Sudoku & puzzles
  • Art

Currently learning

What I'm working on right now

A snapshot, not a scoreboard. No percentages — only what is actually open in front of me.

Building right now

A private website for a driving school (no preview here, by the client's request), and continued work on the transit app with the Spring team.

Where I want to go next

  • An early-stage cancer-detection neural network — an idea I want to explore seriously, not something I have built yet.
  • Deeper work in theoretical physics and mathematics.
  • Getting properly good at C, and at writing secure code by default.

On the desk

  • C# / .NET Desktop apps and web backends
  • Frontend Modern tooling, React, accessibility
  • Python Automation and small tools
  • C Just starting — the fundamentals
  • Lua Picked up while exploring game scripting
  • Cybersecurity Recon, scanning and secure design

Contact

Let's compare notes

If you work on something interesting — or you're also learning and want to trade ideas — I'm happy to talk. The quickest way to reach me is LinkedIn.

Message me on LinkedIn See my code on GitHub

Open to questions, collaborations and honest feedback.