Switch to English version?

Yes

Переключитись на українську версію?

Так

Переключиться на русскую версию?

Да

Przełączyć się na polską wersję?

Tak

Login
Registration
- Welcome to Freelancehunt
  
  Work risk-free, saving time and money
  
  Login Registration

Heorhii Zhukov

Offer Heorhii work on your next project.

Kyiv, Ukraine

1 hour 51 minutes back

available for hire

24 Safes completed

9 days 1 hour back

14 clients

on the service 9 months 26 days

selenium
chat-bot
discord bot
telegram python bot
python

Rating

Successful projects

Average rating

Rating

Bot Development

50 place out of 1969

Data Parsing

70 place out of 786

Bot Development

Data Parsing

Python

AI & Machine Learning

CV

For over five years, I worked as a Senior C++ Developer at TechCore Systems, where I was responsible for creating and optimizing drivers for industrial equipment.

One of my tasks was to speed up signal processing in a driver. After optimizing the code and threads, I managed to reduce latency from 14 to 6 milliseconds, which significantly improved system stability.

After the war began, I moved to Kyiv. During this period, I started actively learning Python and focused on automation and neural technologies.

I am currently working as a Senior Python Developer at CodeLink Technologies, where I develop systems that automate routine processes, increase business efficiency, and simplify data management.

In my free time, I work as a freelancer on international platforms, developing parsers, Telegram bots, and websites.

Skills and abilities

Programming

AI & Machine Learning
from 22 USD for project
Bot Development
from 22 USD for project
C & C++
from 223 USD for project
Data Parsing
from 22 USD for project
Desktop Apps
from 22 USD for project
Embedded Systems & Microcontrollers
from 67 USD for project
HTML & CSS
from 22 USD for project
Python
from 22 USD for project
Web Programming
from 22 USD for project

Portfolio

Multilingual recruiting landing page

HTML & CSS

Developed a landing page in three languages (RU/EN/UA) for a streaming agency with 4000+ active models. The page includes information about earnings ($0.80+/min), onboarding stages, testimonials, FAQ, and an application form. Dark design, mobile-first layout. The landing page is integrated with CRM — applications are processed automatically after the form is submitted.

Stack: HTML, CSS, JavaScript
SaaS platform for tutors — TutorFlow

Web Programming

Built a SaaS product that automates the tutor's routine. The system sends reminders about lessons to students via Telegram, records payments, and after each class sends an AI report on progress to parents. Free plan for up to 15 students, no credit card required.
CRM dashboard for a streaming agency

Web Programming

Developed a CRM system for an agency managing over 4000 models across several sub-agencies. The dashboard displays daily and monthly earnings in coins and USD, online hours, risk zones, and the history of Split payments. Implemented a multi-agency view with live data updates, an administrative panel, and a role-based access model. The project is live in production.

Stack: Python, FastAPI, REST API, database design.
Setting up a WireGuard VPN server with a web panel

Python

Deployed a VPN server on Ubuntu 22.04 (Netherlands) from scratch. Configured WireGuard — a modern fast VPN protocol. Developed a lightweight web panel using Flask, through which you can connect to the VPN in 30 seconds — without a console and technical knowledge.

What has been done:
- Installation and configuration of WireGuard (keys, traffic routing, NAT)
… - Web panel with a QR code for mobile devices and a download button for the PC config
- Auto-start of VPN and panel via systemd
- Connection instructions for Android, iOS, Arch/Artix Linux

Stack: Python, Flask, WireGuard, systemd, Ubuntu 22.04
AI Manager

AI & Machine Learning

AI Manager solves a simple problem: when you can't or don't want to be constantly available, people are left without answers. Not because you don't want to respond — just busy, sleeping, not looking at your phone. Regular auto-responders provide robotic, template text that is immediately obvious. This bot works differently.

The system learns how exactly you write: your tone, message length, vocabulary, common phrases, and reactions. Based on this, a behavioral profile is built. When a message comes into the selected chat, the bot responds in your style: briefly if the message is short; with swearing if that's how you communicate; with the same pauses and intonations that are characteristic of you. From the outside, it looks like you yourself.

… In addition to the auto-responder, there are several separate tools. The news digest collects everything that has happened in the monitored channels since the last request and provides a clean summary without ads and junk. The "Ask" section allows you to ask a question and get an answer based only on what is in the indexed chats. Contacts automatically build cards for each interlocutor — who they are, what was discussed, what remains unresolved. Voice messages are transcribed via Whisper, and the bot responds to them in text.

Everything stored in the database is encrypted. The Telegram session, chat history, contact profiles — all under Fernet encryption. No .session file on disk. The bot does not respond to other bots and does not react to spam.

Stack: Python 3.11, Telethon for user bot, aiogram 3 for management bot, OpenAI API (gpt-4o-mini and whisper-1), aiosqlite, cryptography.
Parser prom.ua

Data Parsing

The script addresses the task of transforming the ordinary web catalog Prom.ua into a structured data source suitable for automation and analysis.

From a technical standpoint, it solves the problem of the lack of a public full API: instead, it uses a combination of HTML parsing and an internal GraphQL query (`/graphql`), which the site itself uses to load data on the product page.

… It combines two levels of data: static HTML (product list, basic parameters) and dynamic GraphQL (delivery, payment, availability, regions, seller business logic). This eliminates the fragmentation of information, where part of the data is available only in the interface and part only through the API.

It also addresses the problem of scaling access to data: instead of manually opening pages, it implements automatic crawling of categories with pagination and sequential processing of products.

From an engineering perspective, it ensures normalization and unification of data: different response formats (HTML + JSON GraphQL) are brought to a unified structure and saved to a file.

Additionally, it serves as a monitoring tool: it allows tracking prices, availability, delivery conditions, and changes from sellers in an automated manner.

As a result, it is not just a parser, but a mini ETL pipeline (extract → transform → load) that transforms the web interface of the marketplace into a database suitable for analysis and automation.
NeuroLabyrinth - a platform for parallel learning of agents

AI & Machine Learning

This is a high-performance reinforcement learning system built around the multi-agent environment Maze RL, optimized for real computational loads and scaling.

The project implements a fully vectorized simulation, where dozens of agents are trained simultaneously in the same environment. The architecture is specifically designed to move away from the classic single-agent approach and replace it with parallel learning in a shared environment, which radically increases the efficiency of computational resource usage.

… Each agent operates independently, but within a unified environment, allowing for the modeling of competitive and collective behavior simultaneously. The system supports batch processing of observations and actions, where all agents undergo a forward pass in a single operation, without N separate model calls. This provides a significant performance boost and makes the system suitable for scaling to hundreds of agents.

The environment is built on procedural generation of complex mazes with controlled structural entropy: cycles, traps, dead ends, false paths, and narrow corridors. This creates a rich learning space where the agent cannot rely on trivial strategies and is forced to develop a robust navigation policy.

The system supports dynamic visualization on Pygame, where all agents are displayed in real-time, including their positions, progress, and aggregated learning statistics. If necessary, the visualization can be turned off, and the system switches to a high-speed headless mode, achieving thousands of agent-steps per second on the CPU.

Training is based on a DQN architecture with a replay buffer, target network, and epsilon decay, adapted for multi-agent mode. Instead of the classic episodic cycle, a streaming step training is used, where model updates occur continuously as experience is received from all agents.

As a result, it is a system that serves both as a research platform and an engineering tool: it demonstrates the behavior of complex RL agents under conditions of dense parallelization, allows for testing scaling strategies, and visually observes collective learning in real-time.

Essentially, it is not just an agent simulator, but a full-fledged environment for developing and stress-testing reinforcement learning algorithms in multi-agent scenarios with high interaction density.
Gemini Watermark Remover

Python

A Python tool that automatically removes the watermark ✦ from images generated by Google Gemini. The problem is that Gemini places a small four-pointed symbol in the bottom right corner of each generated image, and it needs to be removed manually. This tool finds and cleanly removes the mark with one command, processing a folder of images all at once. It works completely offline, does not require API keys or paid services. Supports JPG, PNG, WEBP, and other popular formats.
GitHub Repository Finder

Python

I wrote a console tool in Python for searching GitHub repositories based on semantic descriptions. The program solves a simple problem — the standard GitHub search only looks for exact words in the title, while this tool understands the meaning of the query.

The user enters a short description in 2–3 words, such as "data visualization" or "telegram bot," and the program finds the most suitable repositories by itself. Under the hood, TF-IDF and Cosine Similarity algorithms work — they convert text into numerical vectors and compare them with each other.

… The project is written in pure Python without third-party libraries. The code is divided into modules with a separation of responsibilities: a separate file for algorithms, for working with the API, for error handling with logging. The GitHub token is optional — it works without it, but is faster with it.
Absolute Stress-Screen

Python

Immortal Screenshot + Stress Test is a lightweight and efficient Windows application that automatically performs regular screen screenshot captures at a specified interval while simultaneously conducting intensive stress tests on the CPU and RAM.

The application automatically creates daily folders with the current date in the selected directories, takes full-screen screenshots with an exact timestamp, and saves them in PNG format. For each screenshot, a corresponding text file is created with the content "immortal."

… The program ensures maximum CPU load by performing heavy mathematical calculations on all available cores, as well as maximum aggressive load on the RAM by continuously allocating large blocks without freeing them in a true crash-test mode.

Technology stack: Python 3, MSS, Pillow, Schedule, Tkinter, Multiprocessing, Threading.

The application runs completely in the background, requires no complex setup, and allows for easy addition of auto-start with Windows.

Important warning: the tool contains a true stress test for RAM without any limitations. Usage may lead to significant system slowdown, freezing, or forced closure of other applications. It is recommended to use only for testing purposes on a dedicated machine.

Repository: https://github.com/fedyaqq34356/Immortal-Screenshot-Stress-Test.git
backup-dots

Python

backup-dots
Script for automatic backup of dotfiles and configs on GitHub. Synchronizes 30+ folders from ~/.config/, detects changes, commits with a timestamp, and pushes. Also without terminal.
https://github.com/fedyaqq34356/backup-dots
update-mirrors

Python

update-mirrors
a script for Arch Linux that automatically finds the fastest mirrors through Reflector and updates Pacman databases. Works without a terminal, sends notifications to the desktop.
https://github.com/fedyaqq34356/update-mirrors
Bash script for cleaning the Linux system with notifications

Python

Light Bash script for automatic disk cleanup on Linux. It removes package cache, orphaned dependencies, system logs, and user caches. It sends desktop notifications with a progress bar during the process and a final report detailing how much space has been freed in each partition.

What the script can do:
- Backs up the list of installed packages before cleanup
… - Completely clears the pacman cache and AUR helpers (yay, paru)
- Removes orphaned dependencies
- Trims the pacman log
- Cleans systemd logs (keeps the last 50 MB)
- Deletes safe directories of user cache (thumbnails, mesa shaders, fontconfig, etc.)
- Clears pip cache
- Sends a notification with the percentage of completion at each step
- At the end, sends a final notification with a breakdown by partitions

Stack:
Bash, pacman, systemd, notify-send, pip

For whom:
Users of Arch Linux and other distributions on systemd. Works with Hyprland, KDE, GNOME, and any other environment that supports notify-send.

GitHub: https://github.com/fedyaqq34356/arch-cleanup.git
Telegram bot with the character of a drunken 45-year-old man.

Bot Development

Telegram bot with the character of a 45-year-old drunk man. Responds briefly and genuinely — with profanity, fragmented thoughts, and random philosophy. Maintains the dialogue history, can generate images on request, and comments on them in his own style. Works completely for free without paid APIs.
Stack: Python, aiogram 3, g4f, aiohttp, systemd

GitHub: https://github.com/fedyaqq34356/DrunkenBot.git
Tool for instant text recognition from the screen

Python

A tool for instant text recognition from the screen. You select an area with the mouse — the text is automatically copied to the clipboard. Works completely offline, without clouds and third-party services.

Supports Russian, Ukrainian, and English languages. Compatible with any Linux distribution on Wayland — Arch, Ubuntu, Fedora, openSUSE.

… The source code is open, available on GitHub: https://github.com/fedyaqq34356/screen-ocr
Telegram bot for automating the resale of Telegram groups

Telegram bot for automating the resale of Telegram groups

A system of two parallel working bots — a sales bot and a support bot. Written in Python with aiogram 3.x, Telethon, and async SQLite. Fully asynchronous architecture, both bots are launched in one process via asyncio.gather.

… Sales bot. The buyer browses the catalog of groups with pagination, each group is displayed with its name and age since creation. The user selects a group by number, the request is saved in the database without notifications. The administrator opens the queue of requests upon request — sees the buyer's data and buttons to confirm or decline. Upon confirmation, the Telethon account automatically adds the buyer to the group, grants them full administrator rights, and deletes the chat history of the account in that group.

Support bot. A separate bot with its own token, works in parallel. A live chat is implemented between the buyer and the administrator via relay: messages from both sides are forwarded with inline buttons to reply, the dialogue continues until the buyer presses the end conversation button on the reply keyboard.

Integration with Telethon. Sessions are stored in files, each account is assigned a random device model and application version. Automatic migration of accounts from data.json is implemented upon first launch. If the session is already authorized — the account is added without a code request.

Gift mechanism. The user can transfer rights to an external group to the Telethon account. The bot detects the transfer through the UpdateChannelParticipant event and automatically adds the group to the catalog for sale.

Stack: Python 3.10+, aiogram 3.7, Telethon 1.34, aiosqlite, python-dotenv.
Telegram bot for automatic sale of Telegram stars/premium

Bot Development

The bot automates the sale of Telegram stars and Telegram Premium subscriptions. The user selects the product, specifies the recipient, pays via card, and sends a screenshot of the confirmation. The administrator receives a notification, checks the payment, and with one click confirms or rejects the order — after which the bot automatically delivers the product via an external API.

Stack: Python, aiogram 3, aiosqlite, aiohttp

… Implemented functionality:
Onboarding a new user: language selection (RU/EN/UK), acceptance of the agreement, currency selection (USD/UAH/EUR). Upon repeated launches, the setup is not shown again.

Purchase of stars: the user enters the quantity (from 50 to 1,000,000), specifies the recipient's username, and receives the amount to be paid in their currency with the current exchange rate. After payment, they send a screenshot — the order goes for verification to the administrator.
Purchase of Telegram Premium: selection of the term (3/6/12 months) with fixed prices, similar payment flow.

Administrator panel: viewing all users with pagination, mass mailing, managing the price directly from the chat, confirming and rejecting orders with profit displayed for each.

Product delivery: after confirmation, the bot sends the order to the Wizard API, updates the user's statistics, and offers to leave a review. The review is published in a separate channel.

Multi-currency: exchange rates for USD/UAH/EUR are pulled from an external API with caching for one hour, fallback to fixed values when unavailable.
Multifunctional Telegram bot for promotion automation

Bot Development

Owners of Telegram channels face one problem — algorithms promote content with high engagement. Posts with reactions get recommended and appear more authoritative. Manual boosting through third-party services is expensive and gets flagged by Telegram as bot activity. This project automates promotion through real authorized accounts with full control over logic and frequency.

The system places reactions on new posts on behalf of a pool of real accounts, parses content from third-party channels, and publishes on a schedule, converts videos into circles, manages delayed publications, and processes payments. All through a Telegram bot with support for a custom bot for each user.

… The project is based on aiogram 3 and asyncio. Simultaneously, the main bot's polling, custom bots' polling, reaction worker, scheduler, and payment poller operate — all in one event loop through asyncio.gather without threads and processes.

The key dependency is Telethon, a library for MTProto. Unlike the Bot API, it allows actions on behalf of a regular user — placing reactions from real accounts and reading channels without admin rights. Accounts are authorized through the bot with SMS and 2FA, and the session is saved to a file. For anti-detection, each client receives random device_model, system_version, and app_version.

The reaction worker listens for new posts through events.NewMessage, waits for a specified interval, then random accounts from the pool place reactions through SendReactionRequest with random pauses between calls.

Custom bots: the user provides a token, the system starts polling through asyncio.create_task and feeds updates into the same Dispatcher. All handler code works the same, outgoing messages come from the user's bot.

The database is SQLite through aiosqlite with WAL mode for parallel operations without locks. Payment integration with CryptoPay and Telegram Stars. CryptoPay: an invoice is created, the poller checks the status every 30 seconds, and upon payment, activates the subscription automatically.

The parser iterates through history via iter_messages, downloads media, re-saves through Bot API to obtain file_id, and queues posts with a time offset from the original date. FSM manages multi-step dialogues, and middleware pulls the user's language from the database before each handler.
45 USD
Telegram bot for selling stars — major update

Bot Development

What was before the update:
The bot allowed selection only from a fixed set of ready-made star packages — 15 options divided into two pages with pagination (13, 21, 26, 50, 60, 70, 80, 90, 100, 200, 300, 400, 500, 1000, 10000 stars). Each package had a manually set price without any pricing logic. There was no user profile at all. There was no referral system. Reviews were published only if the user left them themselves. Payment was accepted by card and TON. The database stored only user_id and reviews without any statistics.

What has changed:
… The pricing system has been completely redesigned — a tiered model has been introduced: the price per star automatically decreases as the quantity increases. Instead of 15 fixed packages, there are now 4 quick options and free input from 50 to 100,000 stars with automatic cost calculation and a confirmation screen before payment. Pagination has been removed as unnecessary.
A full user profile has been added with purchase history: total number of stars purchased, total top-ups in hryvnias, referral balance.
A referral system has been implemented: each user receives a unique link, and for each purchase made by a referral, 1% of the purchased stars is credited to the referral balance. The balance can be withdrawn through a request to the administrator.
The review system has been automated: if the user does not leave a review within an hour after purchase, an anonymous entry is automatically published in the channel. This allows maintaining the activity of the review channel without user participation. A separate type of review has been added for displaying referral stars.
TON payment has been removed, the process simplified — now only card payment with the submission of a screenshot of the receipt.
The database has been expanded: fields username, full_name, total_stars, total_uah, referral_balance, referred_by have been added. Automatic migration of existing users has been set up upon launch.

Stack: Python, aiogram 3, SQLite, FSM for multi-step dialogues.
Update Telegram bot for selling Telegram stars and Premium

Bot Development

Project: Telegram bot for selling Telegram Stars and Premium subscriptions
Main functionality: Automated sale of virtual currency Telegram Stars and premium subscriptions through a Telegram bot with payment system integration (Ukrainian bank cards, TON cryptocurrency), a review system, and an admin panel.

What was initially: Users could only choose from a fixed set of pre-installed star packages with different prices (50 stars for 48 UAH, 100 for 85 UAH, 200 for 160 UAH, etc.). The list of 15 different packages was split into two pages with pagination buttons. Each package had an individual price without a single price list. The user could not buy an arbitrary number of stars - only ready-made options.
…
What has been done now: A flexible star purchasing system has been implemented with two operating modes. The first mode - quick packages (13, 21, 26, 50 stars) with automatic cost calculation at a unified rate of 0.84 UAH per star, where the 50-star package is marked as a hot offer with a fire icon. The second mode - free input, where the user specifies the desired number of stars from 1 to 100000, after which the bot automatically calculates the cost and shows detailed order confirmation before payment. A "Enter your amount" button has been added to the star package selection menu.

Technical changes: The pricing system has been centralized through the constant STAR_PRICE_PER_UNIT, allowing the price per star to be changed throughout the system by editing a single number. A new FSM state has been created to handle user input for the number of stars. Validation has been implemented with a minimum check of 1 star and a maximum of 100000 stars. An order confirmation function has been added with a preview of the quantity and cost before proceeding to payment. The structure of the STAR_PRICES dictionary has been simplified by removing many fixed packages and adding dynamic calculation. Code readability has been improved by introducing a named constant for the price instead of hard-coded values. Pagination of the star menu has been removed, as now only 4 quick packages are displayed plus a button for arbitrary input.

Result: Increased flexibility of the system for users, simplified price administration, improved user experience due to the ability to purchase the exact required number of stars without overpaying for rounding to the nearest package.
KeyLogger

Python

KeyLogger

A lightweight cross-platform keylogger written in Python. The program records keystrokes and saves them to text files in real-time. It works on Linux of any distribution and Windows 10/11.

… MAIN FEATURES

The program saves each keystroke instantly to disk. Files are created automatically with numbering if the previous log already exists. Full recording of all languages is supported including Russian, Ukrainian, Chinese, Arabic, and others thanks to UTF-8 encoding. Special keys are handled in a special way - space, enter, and tab are recorded as real characters while others like backspace and delete are written in square brackets. The program exits correctly when the ESC key is pressed.

WORKING LOGIC

Upon the first launch, a file named txt.txt is created. If such a file already exists, txt1.txt is created, and so on. Numbering continues automatically. Letters, numbers, and symbols are recorded as they are. Space, enter, and tab are recorded as actual characters. Backspace and delete are written as [BS] and [DEL]. Modifier keys Shift, Ctrl, and Alt are not recorded. Function keys are written in square brackets, for example, [f1] or [up]. The ESC key stops the program.

INSTALLATION AND USAGE

Clone the repository and navigate to the project folder. Create a virtual environment with the command python -m venv venv. On Windows, activate it via venv\Scripts\activate, and on Linux via source venv/bin/activate. Install dependencies with the command pip install -r requirements.txt. Run the program with the command python main.py. On Linux, it may require running with administrator rights via sudo.

PROJECT STRUCTURE

The file main.py is the entry point and contains the keyboard listener. The logic for file management and data recording is implemented in logger.py. The file config.py contains program settings including the file name and encoding. Dependencies are listed in requirements.txt.

TROUBLESHOOTING

If an error about the pynput module appears when running with sudo, run it with the full path to python from the virtual environment or install pynput in the system Python. If keys are not recorded on Linux, add the user to the input group with the command sudo usermod -aG input $USER. If unreadable text appears, ensure that ENCODING utf-8 is set in config.py.

TECHNICAL DETAILS

The project requires Python 3.8 or higher and the pynput library. The project is distributed under the GNU General Public License v3.0. Created for educational purposes.

https://github.com/fedyaqq34356/KeyLogger.git
Telegram bot for conducting giveaways

Bot Development

Telegram bot for conducting giveaways

A professional bot for organizing giveaways and contests in Telegram channels with advanced automation capabilities.

… Main features:
The bot allows channel administrators to create and manage giveaways with customizable participation conditions. It supports two participation modes: manual, where users click a button to participate, and automatic, where all channel members are added instantly. Administrators can require subscription to multiple channels before participation.

Scheduling and management:
Giveaways can be published immediately or scheduled for a specific date and time. The completion conditions are flexible - either upon reaching the target number of participants or at a specified time. The bot automatically selects random winners and announces the results in the channel.

Advanced functionality:
The built-in participant management system allows adding or removing users manually. Active giveaways can be edited in real-time including text, button labels, number of winners, and end time. The mailing feature sends personalized notifications about the giveaway to all channel participants.

Technical implementation:
Created in Python using the aiogram framework for bot functionality and Telethon for advanced Telegram API capabilities. Uses an SQLite database for data storage. Supports multiple time zones and includes automatic schedulers for time-based giveaways. Session management ensures continuous authentication for automated functions.

Automation:
Automatic participant scanning uses the Telegram API to collect all channel members except administrators and bots. The mailing system sends messages to channel participants with flood protection and automatic retry logic. Background schedulers check completed giveaways and scheduled publications every 30 seconds.

Security:
Administrator access is controlled by user ID verification. Environment variables for sensitive data. Session files are stored securely. Subscription verification before participation. Rate limiting is implemented to prevent abuse of the Telegram API.

Technology stack:
Python 3.8+, aiogram for Telegram Bot API, Telethon for MTProto, SQLite for the database, pytz for timezone handling. Modular architecture with separate handlers for different bot functions.

Installation:
Simple setup process with a virtual environment and pip dependencies. Configuration through environment variables. No complex deployment requirements - works on any server that supports Python.

Repository:
https://github.com/fedyaqq34356/Random-bot.git

Questions and support:
https://github.com/fedyaqq34356/Random-bot/issues

The bot is fully functional, tested, and ready for deployment. Open source with clean, documented code structure.
29 USD
Version 2.0 — Major update of the Valencia bot

Bot Development

Version 2.0 — Major update of the Valencia bot

Multilingualism. The bot is fully translated into three languages: Russian, Ukrainian, and English. Now, upon first launch, the user selects a language by entering ru, uk, or en, and all subsequent messages — instructions, questions, notifications — come in the chosen language. The greeting upon registration, instructions for downloading the app, error messages, group buttons — everything is adapted for each of the three languages.

… Large knowledge base on Halo. The AI processor has a huge local knowledge base HALO_TRAINING_KNOWLEDGE in all three languages. It contains detailed instructions on hunting, multi-beams, profile setup, posting, launching regular and premium broadcasts, content rules, dislike system, auto-messages, tasks. Previously, the bot relied only on the FAQ from the database and external AI — now most operational questions are resolved instantly without a request to AI.

Dislike coefficient calculator. A separate function has been added that analyzes user messages, extracts dislike and like numbers, calculates the coefficient using the formula dislikes/likes, and immediately indicates whether it is normal or in a danger zone. If the user asks for a calculation but hasn't provided numbers, the bot asks for clarification of data for the last 30 days.

Smart blocking of operational questions. Users who have not yet joined groups now do not have access to operational topics — hunting, broadcasts, dislikes, posts. They receive a message stating that access to training will open after registration is completed. An exception is made for questions about earnings — this information is available to everyone during the introduction to the project.

Pagination in the user list. Previously, all users were displayed in one list, which became inconvenient as the database grew. Now, pagination has been implemented with 10 people per page and page-switching buttons. This affected the sections for viewing conversations, writing to users, and deleting conversations.

Deleting conversations and hiding users. The administrator can now delete the conversation of a specific user and hide them from the list. Hidden users automatically reappear as soon as they send a new message. A background task auto_hide has also been added — it automatically hides inactive users: those who remain silent for more than an hour during registration, and registered users who have not written for more than 8 hours.

Managing group links. A new section has appeared in the admin panel where links to the training group and chat with girls can be changed directly through the bot interface without editing code or the database.

Multilingual greeting. Now each of the three languages has its own separate greeting message in the database. The administrator can edit each one independently — the language is selected, and then new text is entered.

Translation of training materials. When the command /startanal is launched, all texts, audio, and video captured from the group are now automatically translated into Ukrainian and English via AI. The translations are saved in the database separately and used when responding to users in the corresponding language.

Language detector. A language_detector module has been added with two functions: one determines the language by selection keywords (ru/uk/en), the other automatically identifies the language of any text based on Ukrainian specific characters, characteristic words, and the ratio of Cyrillic to Latin.

Protection against repeated screenshots. Registered users sending photos now receive a message stating that registration has already been completed, instead of re-launching the screenshot processor.

Determining the intent to register. A list of patterns REGISTRATION_INTENT_PATTERNS has been added — if a registered user writes something like "I want to register" or "sign up," the bot politely reminds them that registration has already been completed.
67 USD
Telegram Privacy & QR Bot

Bot Development

Project Description
A two-bot system for Telegram with user authorization features through Telethon, automatic group moderation, and scanning of QR codes/barcodes. It includes userbot functionality for automatically deleting unwanted dialogues with group participants while maintaining communication with administrators.

What Problems It Solves
… Administrators of closed Telegram communities face spam in personal messages from group participants, rule violations (posting contacts in the chat), and the need to manually check QR codes and barcodes from participants. Every day, they have to delete dozens of spam messages, track violators, and verify access codes. The system automates all these tasks - protects administrators' personal messages, moderates content in the group, and processes QR/barcodes without human involvement.

Monetization
Selling access to the bot for closed communities and paid channels - from 300 UAH/month for basic functionality to 1500 UAH/month for a full package tailored to the project's specifics. White-label solutions for large communities - from 5000 UAH for a one-time installation. Additional income from implementation in corporate Telegram groups of companies for controlling employee communications.

What Are the Benefits
Complete protection of the administrator's personal account from spam from group participants. Automatic content moderation saves 10-15 hours a week on manual message verification. Instant processing of QR codes instead of using third-party applications. Account privacy settings are automatically configured upon authorization. A flexible rule system for filtering unwanted content in the group.

Technical Stack
Python, Aiogram for bot API, Telethon for MTProto API and userbot functionality, OpenCV + pyzbar for QR/barcode recognition, FSM (Finite State Machine) for managing authorization states.

GitHub: https://github.com/fedyaqq34356/TelethonQRPrivacyBot.git
42 USD
Avto.pro Parser

Project Description
Telegram bot for automating the search for auto parts on the Avto.pro platform. The system parses offers by article numbers, extracts information about prices, delivery times, and suppliers, and saves the results in Google Sheets. It operates in multithreaded mode with three workers to speed up the processing of large lists of article numbers.

What Problems It Solves
… Auto service owners and parts sellers spend hours manually searching for offers for each article number. They have to open dozens of pages, compare prices, and copy data into spreadsheets. When working with 50-100 items a day, this takes 3-4 hours of pure time. The bot automates the entire process - you upload a list of article numbers to Google Sheets, start parsing, and receive a structured table with all offers, prices, and delivery times.

Monetization
Subscription model for auto services and parts stores - from 500 UAH/month for the basic rate (up to 1000 article numbers per day) to 2000 UAH/month for premium (unlimited + priority processing). Ready-made supplier databases with current prices can be sold. Additional income from integration with CRM systems in the auto business - one-time setup from 3000 UAH.

What Are the Benefits
Saving 15-20 hours of work per week on searching and comparing prices. Automatic updating of price lists without human involvement. The ability to quickly check availability and prices from all suppliers simultaneously. Reducing the likelihood of errors during manual data entry. Competitive advantage due to the speed of processing customer orders.

Technical Stack
Python, Selenium WebDriver for browser automation, Aiogram for the Telegram bot, Google Sheets API (gspread) for working with spreadsheets, multithreading via ThreadPoolExecutor.

GitHub: https://github.com/fedyaqq34356/AvtoProParser.git
45 USD
AI Recruitment Telegram Bot

AI & Machine Learning

Understood, my mistake. Let's make it denser: without unnecessary empty lines, but **not as one continuous block**, rather in normal paragraphs. Below — neatly and compactly.

---

… Advanced Telegram bot for recruiting agencies working with streaming platforms. The bot includes AI responses, automatic application processing, photo verification, OCR for data extraction from documents, and a full admin panel.
Note: the bot is specifically designed for agencies recruiting models for the streaming application Halo.

Functionality.

User capabilities.
AI chat with intelligent responses based on g4f (GPT-4 Free) with automatic FAQ matching. Smart handling of photos with support for single and grouped uploads. Step-by-step application process. Feedback system with an image gallery for social trust. Multilingual interface in Russian and Ukrainian with a natural communication style.

Administrator capabilities.
Control panel with real-time statistics and AI performance metrics. Application verification with the ability to approve or reject. Monitoring all user dialogues and exporting chat history. Automatic AI training on administrator responses. Flexible customization of welcome messages and solution templates. Management of prohibited topics. Export of logs for debugging.

Technical capabilities.
OCR integration for automatic data extraction from document screenshots using Tesseract. FSM for controlling user scenarios. SQLite with aiosqlite for asynchronous operation. AI confidence assessment with automatic escalation to the administrator. Proper handling of media groups. Mechanism for repeated AI requests with fallback.

Requirements.
Python 3.8 or higher. Telegram bot token. Installed Tesseract OCR.

Installation.
Cloning the repository and navigating to the project folder. Creating and activating a virtual environment. Installing dependencies from requirements.txt. Installing Tesseract OCR (Windows — via installer, macOS — brew install tesseract, Linux — apt-get install tesseract-ocr).

Environment setup.
Creating a .env file in the root of the project specifying BOT_TOKEN and ADMIN_ID.

Additional files.
Goods folder for review images. Images folder with halo_download.jpg file for a screenshot example.

Running.
Running the main.py file. Upon startup, the bot initializes the database, loads FAQ and settings, starts receiving messages, and logging. Stopping via Ctrl+C.

Usage for candidates.
Starting via /start. Communicating with AI. Sending 2–3 photos. Answering questions about schedule and experience. Waiting for verification. Receiving instructions upon approval. Sending a screenshot with a document. Activating the account on the next working day.

Usage for administrator.
Command /admin to enter the admin panel. Viewing statistics. Checking applications. Answering questions forwarded to AI. Exporting conversations and logs. Customizing messages and settings.

Configuration.
Through config.py, you can change the number of photos, AI confidence threshold, prohibited topics, and universal responses. SYSTEM_PROMPT allows managing the style and logic of AI.

Database.
Tables used include users, messages, photos, applications, faq, ai_learning, settings, forbidden_topics, and pending_questions.

AI system.
A multi-level approach is used: direct FAQ matching, context building, confidence assessment, automatic escalation, and a training system based on administrator responses. g4f (GPT-4 Free) is used as the AI provider.

Adaptation.
The bot can be adapted to other niches by changing texts, scenarios, questions, languages, and adding new functionality.

License.
The project is distributed under the GNU GPL license.

Support.
Questions and suggestions are accepted via Issues on GitHub.

Repository link:
https://github.com/fedyaqq34356/HR-AI-Telegram-Bot.git
Image Generator

AI & Machine Learning

Image Generator G4F

A Python-based tool for generating images using the g4f library, allowing the creation of AI-generated images based on text prompts.

… !Example: Image generated from the prompt "apple"

Capabilities

- Generation of high-quality images using the flux-pro model
- Automatic saving to a local directory
- Simple command-line interface
- Modular code structure
- Use of any available g4f provider

Requirements

- Python 3.10+
- Internet connection

Installation
1. Clone the repository:
git clone https://github.com/YOUR_USERNAME/g4f-image-generator.git
cd g4f-image-generator

2. Create a virtual environment:
python -m venv venv

3. Activate the virtual environment:
Windows:
venv\Scripts\activate

macOS/Linux:
source venv/bin/activate

4. Install dependencies:
pip install -r requirements.txt

Usage
Run the generator:
python main.py

Enter the prompt when prompted:
Enter prompt: beautiful sunset over the ocean
Generated images will be saved in the `generated_images/` directory with timestamps.
Configuration
Edit `config.py` to customize parameters:

OUTPUT_DIR = "generated_images" # Output directory
DEFAULT_MODEL = "flux-pro" # AI model used
TIMESTAMP_FORMAT = "%Y%m%d_%H%M%S" # Timestamp format in filename

Project Structure

image-generator/
├── main.py # Entry point and user interface
├── generator.py # Image generation logic
├── file_handler.py # File saving functionality
├── config.py # Configuration settings
├── requirements.txt # Python dependencies
├── generated_images/ # Output directory (created automatically)
└── README.md # This file

How It Works
1. Input: The user enters a text prompt
2. Generation: The g4f client generates an image using the flux-pro model
3. Download: The image URL is obtained and downloaded
4. Saving: The image is saved with a timestamp in the `generated_images/` directory

Dependencies

- `g4f` - Free GPT-4 API client with image generation support
- `requests` - HTTP library for downloading images

Troubleshooting

Generation failed
- Check your internet connection
- Try again (g4f automatically selects available providers)
- Wait a few seconds and retry

Import errors
- Reinstall dependencies: `pip install -r requirements.txt --upgrade`
- Check Python version: `python --version` (requires 3.10+)

Permission errors
- Ensure you have write permissions to the project directory
- Run with appropriate user permissions

Examples
Basic usage
python main.py
Enter prompt: red apple with water droplets
Output:
Generating: 'red apple with water droplets'...
✓ Saved: generated_images/image_20260109_143052.png
✓ Done!

License
GNU General Public License v3.0 - details in the LICENSE file.

Acknowledgments

- https://github.com/xtekky/gpt4free - GPT4Free library
- The flux-pro model for high-quality image generation

Created using Python and g4f.

GitHub: https://github.com/fedyaqq34356/Image-Generator.git
29 USD
Air Raid Alert Monitor

Python

Air Raid Alert Monitor

Python system for monitoring air alerts in the city of Dnipro, Ukraine, with automatic notifications sent to a Telegram channel in real-time.

… The project is designed to track the current status of alerts through the Alerts in UA API and to instantly inform users about the start and end of alerts. The system allows monitoring of the entire Dnipro region as well as, with additional configuration, all regions of Ukraine.

Functionality:
* Real-time monitoring with customizable intervals
* Automatic notifications in Telegram upon status change of alerts
* Tracking alert statuses: active, partial, no alert
* Logging all events to console and file
* Error handling for API failures
* Ability to customize notification texts

Technology stack:
* Python 3.8+
* Aiogram for integration with Telegram
* Alerts in UA Python API for retrieving alert data
* python-dotenv for managing settings through environment variables
* APScheduler for scheduling periodic checks

Features:
* Ability to monitor multiple regions simultaneously
* Customizable messages with timestamps
* Flexible configuration through .env

Result:
The system allows for prompt informing of users about risks in the region, minimizing delays in response to air alerts.

GitHub: https://github.com/fedyaqq34356/alerts-in-ua.git
45 USD
Trial Period Bot

Bot Development

Trial Period Bot
Telegram bot for managing the trial period (8 days) in the work chat and study group.

Capabilities
… • Tracking the trial period from the moment of joining
• Controlling two chats simultaneously
• Notifications to admins about the trial expiration and user exits
• Checking attendance in both chats
• Admin panel in Telegram
• SQLite database
• Scheduler for regular checks

Installation
git clone https://github.com/yourusername/trial-period-bot.git
cd trial-period-bot
pip install -r requirements.txt

Create .env:
BOT_TOKEN=your_bot_token
WORK_CHAT_ID=-100xxxxxxxxxx
STUDY_GROUP_ID=-100yyyyyyyyyy
TRIAL_MINUTES=11520

Add the first admin:
python add_admin.py YOUR_TELEGRAM_ID

Launch
python main.py

How to get data
• Token: @BotFather → /newbot
• Chat IDs: add @userinfobot to the chat, forward a message
• Your ID: message @userinfobot → /start

Admin panel
Users:
• Export list
• List on trial
• Check attendance
• Remove participant
• Skip trial (early approval)

Administrators:
• Add / remove
• List of admins

Automatic functions
Upon joining:
• Auto-approval of application
• Adding to the database
• Starting the 8-day trial

Monitoring:
• 24 hours before — warning to admins
• Upon completion — notification with "Keep / Kick" buttons

Upon exit:
• Status update
• Notification to admins
• Auto-removal from the study group upon exit from the work chat

Project structure
trial-period-bot/
├── main.py # Entry point
├── config.py # Configuration
├── database.py # Database operations
├── scheduler.py # Task scheduler
├── keyboards.py # Bot keyboards
├── utils.py # Utility functions
├── add_admin.py # Admin addition script
├── handlers/
│ ├── chat_events.py # Chat event handling
│ ├── menu_handlers.py # Menu button handling
│ ├── admin_handlers.py # Admin operations
│ └── callback_handlers.py # Inline buttons
├── .env # Environment variables
├── bot.db # Database (created automatically)
└── requirements.txt # Dependencies

Database
users: telegram_id, name, username, join_date, trial_end_date, status (trial/approved), in_work_chat, in_study_group, notified_one_day
admins: telegram_id

Bot requirements
The bot must be an admin in both chats with permissions to:
• invite users
• ban/kick
• view participants

Deployment
Systemd:
Create /etc/systemd/system/trial-bot.service with ExecStart pointing to main.py, Restart=always
sudo systemctl enable --now trial-bot

Docker:
docker-compose up -d

Troubleshooting
• Bot not responding → token, blocking, internet
• Applications not approved → bot permissions, are applications enabled in the chat
• No notifications → check the admin list and logs

Technologies
Python 3.8+, aiogram 3.15.0, APScheduler 3.10.4, SQLite

License
GNU GPL v3.0

GitHub: https://github.com/yourusername/trial-period-bot
If useful — a star is appreciated.
Telegram message forwarder

Bot Development

Automatic forwarding of messages between Telegram channels in real time. The bot monitors new messages in the source channel and instantly forwards them to the target channel, preserving media files, text formatting, and entities.

Technologies: Python, Telethon, asyncio
Movie Search App

Data Parsing

Movie Search App
Console utility for searching and exporting movie data from the TMDb database.

GitHub Repository: https://github.com/fedyaqq34356/Movie-Search-App
…
Movie Search App is a command-line tool for working with The Movie Database (TMDb) database. The application allows users to search for movies by genres, obtain detailed information, and export data to text files. The project demonstrates skills in working with REST API, modular architecture, data processing, and logging.

Key features:
search by genres, pagination (1 page = 20 movies), data export (titles, descriptions, ratings, providers), regional support, operation logging, error handling.

Technologies used:
Python 3.7+, libraries requests, python-dotenv, logging. TMDb API v3, Bearer token. Development principles: modularity, readability, DRY.

Project structure:
Movie-Search-App/
├── main.py # Entry point and workflow coordination
├── api.py # TMDb API requests and responses
├── config.py # Configuration and environment variables
├── display.py # User interface and input handling
├── export.py # File writing operations
├── logger.py # Logging configuration
├── .env # API credentials (not tracked)
├── requirements.txt # Python dependencies
└── README.md # Documentation

Key decisions:
storing tokens in .env, delays between requests (0.5 sec), gathering all types of providers (subscription, free, ad-supported, rental, purchase), multi-level logging (DEBUG, INFO, WARNING, ERROR), error handling with try-except and retries.

Usage examples:
selecting a genre, entering the number of pages, exporting data to files (titles, descriptions, ratings, providers). The region and language can be changed in the config for correct provider display.

Performance:
1 page (20 movies) ~1 second, 10 pages ~10 seconds, 50 pages ~50 seconds. Reliability and scalability are ensured, supporting up to 500 pages, efficient memory usage, minimal API load.

Possible improvements:
short-term — JSON export, CLI arguments, support for series, genre caching; long-term — GUI, SQLite database, parallel requests, REST API wrapper, unit and integration tests.

Installation:
git clone https://github.com/fedyaqq34356/Movie-Search-App.git
cd Movie-Search-App
pip install -r requirements.txt

Links:
GitHub https://github.com/fedyaqq34356/Movie-Search-App

License:
GNU General Public License v3.0. The project is free to use, modify, and distribute under the terms of GPL v3.
Audio Transcription Tool

Python

Console application in Python for automatic transcription of audio files using the faster-whisper model (optimized implementation of OpenAI Whisper). The project converts the input audio file to text, saves the transcription to a file, and optionally splits it into parts for easier processing of large volumes.

Features:
1) Support for various audio formats through automatic conversion to WAV using FFmpeg (mono, 16 kHz).
… 2) Use of the Whisper large model (customizable in size) for high recognition accuracy.
3) Decoding parameters (beam_size=20, best_of=10, temperature=0.2) for a balance between accuracy and speed.
4) Automatic detection of language and audio duration with logging.
5) Saving the transcription line by line in a text file (.txt).
6) Splitting large text into parts of 500 lines for easier further processing.
7) Detailed logging of all stages to a file for debugging.
8) Minimalist CLI interface: support for specifying the path to audio via command line argument.
9) Full modularity: separate modules for configuration, logging, conversion, transcription, and text splitting.

Example of operation:
When processing a song with vocals, the model attempts to transcribe the text, but due to the musical accompaniment, there are significant inaccuracies: repetitions of phrases, distortions of words, and hallucinations (for example, the repeated line "I'm not trying to be a fool"). This is expected, as Whisper is optimized primarily for speech rather than lyrics in music—the background instrumental part and stylized vocals complicate the task. However, for regular conversational speech, the project demonstrates excellent results: high accuracy even in conditions of strong noise (wind, background sounds, echo), thanks to the robustness of the Whisper model to real acoustic interference.

Technologies:
Python 3, faster-whisper (CTranslate2), FFmpeg (for conversion).

Repository:
https://github.com/fedyaqq34356/audio-to-txt.git
Image in ASCII

Desktop Apps

ASCII Art Converter

Project Description:
A graphical desktop application in Python for converting raster images to ASCII art. The project is implemented as a classic Unix filter: it takes an image and outputs a clean text stream suitable for further processing in pipelines or scripts.
…
Features:
1) Support for a wide range of image formats (PNG, JPG, JPEG, BMP, GIF, WebP).
2) Automatic scaling considering aspect ratio and character height compensation to maintain proportions.
3) Conversion to grayscale with optimal brightness mapping to a limited set of ASCII characters.
4) Real-time preview of ASCII art in a monospaced font.
5) Export of the result as plain text (.txt) or rendered image (.png) with customizable background and text color.
6) Minimalist dark interface on PyQt6 with convenient buttons for opening, copying, and saving.
7) Detailed logging of operations to a file for debugging and auditing.
8) Full modularity: each stage (loading, processing, conversion, rendering, UI) is separated into its own file and easily reusable.

Example of Work:
Converting a portrait retains recognizable facial features and details even with a limited width of 120 characters. In the screenshot: on the left - text preview in the application, on the right - exported PNG image with rendered ASCII.
In the screenshot, the left part is ASCII text in QTextEdit, the right is a rendered png of Joseph Vissarionovich Stalin on the background of the USSR flag.

Technologies:
Python 3, PyQt6 (GUI), Pillow (image processing).

Repository:
https://github.com/fedyaqq34356/image-to-ascii.git
Video Downloader

Data Parsing

Video Downloader
GitHub: https://github.com/fedyaqq34356/video_downloader.git

Description
… Cross-platform video downloader with a modern graphical interface on PyQt6 and yt-dlp.

Key Improvements
1. Redesigned structure - modular architecture with separate handlers for each platform

2. Professional code organization - clear separation of logic based on Unix philosophy principles

3. Increased responsiveness - multithreaded downloading with real-time progress tracking

4. Expanded platform selection - YouTube, TikTok, Instagram, Twitter/X, Twitch, PornHub

5. Improved UX - folder selection for saving, visual progress bar, detailed logging

Technologies
1. Python 3
2. PyQt6 (graphical interface)
3. yt-dlp (downloading engine)
4. Threading (asynchronous operations)

Features
1. Downloading from 6 major platforms
2. Playlist support
3. Thumbnail extraction
4. Customizable save folder
5. Progress monitoring
6. Clean, intuitive interface
OLX.ua Parser

Data Parsing

Asynchronous parser for ads with OLX.ua API.

Technologies: Python 3.7+, aiohttp for asynchronous HTTP requests, openpyxl for data export.

… Key solutions:
- Asynchronous architecture on asyncio for parallel request processing
- Automatic Excel formatting with fixed cell sizes
- Rate limiting through random delays (3-6 seconds) to bypass blocks
- Incremental data saving every 2 pages to prevent data loss

Parsing includes extracting IDs, titles, prices, geolocation, photos, and seller metadata.

#Repository: https://github.com/fedyaqq34356/olx.ua-parser.git
Automated trading bot for Binance

AI & Machine Learning

Binance Trading Bot — an automatic bot for trading on Binance. It operates on a trailing strategy: buys when the price drops and sells when it rises, securing profit.

What it does:

… * monitors selected coins in real-time
* automatically places and updates orders
* buys at the bottom and sells at the peak
* keeps statistics and stores data in MongoDB

What it can do:

* work with multiple coins simultaneously
* use a grid strategy
* connect to TradingView for analysis
* send notifications to Slack
* operate through a web interface (React + WebSocket)
* support three trading modes: based on its own trained model, through OpenAI, or based on signals from Telegram channels

Used at your own risk.
Bot for automatic redemption of gifts

Bot Development

Telegram bot for automatic gift redemption and star management in Telegram. Supports two modes of operation: purchase on behalf of the bot or directly from the user's account.

What it does:

… 1. tracks the appearance of new gifts and instantly redeems the necessary ones
2. supports replenishment and withdrawal of stars
3. sends notifications about new drops in real time
4. logs all actions and automatically restarts in case of errors
5. works through a proxy and supports multithreading
6. has flexible settings for priority and purchase limits
7. can operate in fully automatic or semi-automatic mode

Technical part: aiogram and pyrogram for Telegram API, aiohttp for asynchronous requests, SQLite for data storage, dotenv for configuration, proxy rotation for stability. The bot is fully asynchronous, optimized for high load and instant reaction to drops. The latest test showed excellent results — all necessary gifts were redeemed without delays. Upon request, I can provide screenshots or the bot itself for sale.
Parsing from Rozetka

Data Parsing

Web application on FastAPI that automatically collects product data from rozetka.com.ua. There is an authorization system, a database, and three parsing modes. All results are saved in Excel categorized by categories.

What it does:

… 1. Parses products by search query or category — you insert the link, specify the number of pages, and the bot collects all items in Excel.
2. Collects products from a specific seller with data on groupings and minimum prices.
3. Allows saving and parsing favorite products manually or automatically.

What data it receives:
name, price (old and new), rating, number of reviews, brand, category, popularity, seller, payment methods, warranty, specifications, delivery, number of videos and credits. For sellers, metrics on reviews, prices, and groupings are added.

Technical part:
Selenium — for processing dynamic pages (expanding cards, scrolling, etc.), Cloudscraper — for bypassing Cloudflare, up to 10 threads simultaneously, batches of 60 products, random delays to simulate real actions.

Authorization system:
three levels of access — pending, accepted, and admin. JWT tokens are used (24 hours), passwords are hashed via bcrypt. There is an admin panel, by default a user admin1/admin33 is created.

Export to Excel:
products are automatically distributed by categories into separate sheets, columns are highlighted in color — green (main), orange (delivery), gray (popular features), yellow (rare). If there are no popular features in the category, the sheet is marked with three exclamation marks.

Technology stack:
Python, FastAPI, Selenium, Cloudscraper, BeautifulSoup, OpenPyXL, SQLite, JWT, bcrypt, asyncio, ThreadPoolExecutor.
Desktop Image Upscaler

AI & Machine Learning

What it can do:

* increases images by 2–4 times with natural detail and clear textures
* restores faces without the "plastic" effect
… * supports batch processing, drag&drop, and before/after preview
* uses GPU (auto-detection) or CPU, optimizing memory through tiles
* keeps a progress log and works reliably even with large files

Technical part: PyTorch, OpenCV, PyQt interface. Works on Windows and Linux, fully offline with local weights.

For example: upscaling the first episode of Berserk from 480p to 2K took 1 hour and 28 minutes on a 5070 Ti graphics card.
Program for downloading videos from 11 sources

Data Parsing

An automatic program for downloading videos from 11 sources, including YouTube, Instagram, Twitch, Facebook, TikTok, Twitter, and other streaming services.

What it can do:

… 1. works asynchronously, downloading multiple videos simultaneously
2. supports 11 platforms, including streaming services
3. allows the use of cookies and selecting a browser for authorization
4. extracts not only videos but also:
– audio track separately
– comments under the video
– chat from Twitch streams
– number of likes and title of the video
5. has an interface on PyQt6 with a simple and user-friendly design
Bot on aiogram with its own AI model for dish recognition

Bot Development

Telegram bot on aiogram with its own AI model that recognizes dishes from photographs.

What it does:

… 1. identifies the name and type of dish from the image
2. analyzes the composition and lists the main ingredients
3. calculates approximate calorie content and macronutrients
4. assesses portion size and provides a description of health benefits
5. accepts multiple photos and saves the history of analyses

The bot operates completely autonomously, uses a built-in AI model without external APIs, processes images locally, and generates a structured response in the Telegram chat.
71 USD
AI-Tarot Reader

AI & Machine Learning

I created a web application for a client for Tarot card reading with a modern approach. Ask your question, choose a spread, and let the ancient wisdom of the cards and the power of artificial intelligence provide you with an answer :D

What's inside:

… AI interpretations: A neural network (g4f) analyzes the drawn cards and your question to provide a deep and coherent interpretation of the spread.
Various spreads: From a quick answer with one card to a detailed "Celtic Cross."
Card of the day: Daily prediction for anyone interested.

Technology stack:

- Frontend: React, Tailwind CSS, Framer Motion
- Backend: FastAPI (Python), g4f
- Database: All information about the cards is stored in the code.
Parsing of the maximum product of Rozetka

Data Parsing

Telegram bot for automatic tracking of product stock on Rozetka with data synchronization in Google Sheets.

What it does:

… 1. accepts product links and saves them in the database
2. automatically checks availability and updates stock on a schedule
3. synchronizes data with Google Sheets and exports history
4. supports bulk addition of products (up to 20 at a time)
5. allows manual stock checking and table updating
6. stores check history and generates reports
7. can delete and edit products
8. works in automatic and manual modes
9. logs all operations and restarts on errors

The bot is fully asynchronous, error-resistant, can restart, log all operations, and update data on a schedule.
Bot for automated sale of Telegram stars

Bot Development

Telegram bot for automated sale of stars and Premium subscriptions in Telegram with Split API integration and two payment methods — via TON and bank card.

What it does:

… * sells star packages (50–10000) and Premium subscriptions (3, 6, or 12 months) with automatic cost calculation
* accepts payment via TON Connect or by screenshots of card transfers
* checks subscription status on the channel before starting work
* collects reviews with ratings and publishes them in a separate channel
* has an admin panel with mailing, statistics, and order moderation
* deletes overdue orders after one hour
* verifies the correctness of the recipient's username
* keeps a log of all actions and restarts in case of failures

Technical foundation: aiogram for Telegram API, aiohttp for working with Split API, SQLite for data storage, FSM for step-by-step scenarios, dotenv for configuration. The architecture is fully asynchronous, with callback request handling and user state management.
Telegram bot crypto exchange

Bot Development

Bot on Aiogram for a cryptocurrency exchange. Asynchronous and resilient under load, it uses APIs (CoinGecko, Binance) to obtain current cryptocurrency rates.

The bot allows users to create and track orders, as well as store transaction history in a database.

… A system for connecting an operator is provided — the user can directly contact support to clarify exchange details or resolve issues.

OpenAI is integrated for intelligent responses, prompts, and automatic assistance in chat.

An admin panel is also implemented for managing rates, orders, and users.

Show more 30 jobs

Reviews and compliments on completed projects 24

17 June 181 USD
Website development

Quality

Professionalism

Cost

Contactability

Deadlines

Quickly and professionally developed a website for me.

Vlad Y. | Personal | Response review

12 June 446 USD
Website development

Quality

Professionalism

Cost

Contactability

Deadlines

Fully implemented my terms of reference, the site works great.

Vlad Y. | Personal | Response review

10 June 22 USD
Improvement of the Telegram bot

Quality

Professionalism

Cost

Contactability

Deadlines

As always, the work was done professionally, with quality, and within the agreed deadlines. All tasks were completed quickly and competently, with attention to detail. We are completely satisfied with the collaboration and the results obtained. We recommend as a reliable and responsible specialist.

Denis S. | Personal | Response review

30 May 40 USD
Improve the bot

Quality

Professionalism

Cost

Contactability

Deadlines

Everything went well, I was able to implement my project quickly.

Vlad Y. | Personal | Response review

22 May 89 USD
Improve the bot

Quality

Professionalism

Cost

Contactability

Deadlines

The bot works great, my technical specifications have been fully met.

Vlad Y. | Personal | Response review

11 May 100 USD
Telegram bot with Claude API and Google Drive — personal AI agent

Quality

Professionalism

Cost

Contactability

Deadlines

Everything was done on time and with quality. I recommend.

Andrii Prepelitsya | Safe

| Response review

8 May 35 USD
It is necessary to set up Windows VDS automatically for operation through Windows App / Remote Desktop.

Quality

Professionalism

Cost

Contactability

Deadlines

Quick and professional implementation of the project

Vlad Y. | Safe

| Response review

6 May 94 USD
Development of a Telegram bot for a garbage collection service

Quality

Professionalism

Cost

Contactability

Deadlines

Georgiy created a Telegram bot quickly, efficiently, and taking into account all our wishes. He was always in touch, explained all the details thoroughly, and promptly made corrections.

I definitely recommend him for collaboration! A professional in his field, with whom it is pleasant to work.

Denis S. | Safe

| Response review

6 May 56 USD
Auto Proxy Setup for Windows VDS (SOCKS5)

Quality

Professionalism

Cost

Contactability

Deadlines

I completed the project in one day, everything works super fast.

Vlad Y. | Safe

| Response review

2 May 27 USD
Development of a Python script for local export of Telegram channels (open/closed)

Quality

Professionalism

Cost

Contactability

Deadlines

Everything is great! Fast and just as needed.

Myro Vakulen | Safe

| Response review

10 February 29 USD
Improvement of the bot

Quality

Professionalism

Cost

Contactability

Deadlines

The developer has golden hands, he completed quite a complex task, when I had questions or needed improvements in the bot, he promptly fixed it.

Vlad Y. | Personal | Response review

4 February 29 USD
Data parsing from the website

Quality

Professionalism

Cost

Contactability

Deadlines

The project was completed on time and in full detail.

Roman Rud | Safe

| Response review

30 January 42 USD
Proposal parser from the website

Quality

Professionalism

Cost

Contactability

Deadlines

Wonderful performer! Everything is fast and of high quality! We will order another project and I recommend everyone to collaborate with Georgiy!

Oleksandr Shulhin | Safe

| Response review

29 January 67 USD
Group in Telegram using a bot for solving some tasks

Quality

Professionalism

Cost

Contactability

Deadlines

I recommend for collaboration )

Dmitry Chenkov | Safe

| Response review

22 January 67 USD
Regular data collection from Rozetka by categories (with reviews and photos)

Quality

Professionalism

Cost

Contactability

Deadlines

Everything was done qualitatively and on time. Thank you for the great work!

Mikita Kalenik | Safe

| Response review

20 January 38 USD
Improvement of the Telegram bot (training)

Quality

Professionalism

Cost

Contactability

Deadlines

He performed his task very professionally.
He created the bot completely according to my specifications, responded quickly to my messages, and resolved the issues I encountered.
I definitely recommend him.

Vlad Y. | Personal | Response review

18 January 22 USD
Telegram bot for parsing channels

Quality

Professionalism

Cost

Contactability

Deadlines

Great performer, I will definitely reach out again.

Aleksandr Yefremenko | Safe

| Response review

14 January 29 USD
Set up automatic publication of messages about the start and end of air alerts in the Telegram channel.

Quality

Professionalism

Cost

Contactability

Deadlines

The work is completed, the cooperation went well.

Darya Pashchenko | Safe

| Response review

12 January 45 USD
Development of a Telegram bot with AI for communication with candidates and selection of applications.

Quality

Professionalism

Cost

Contactability

Deadlines

This is not the first time I am working with this freelancer, as always he completed the task requirements with a score of 10 out of 10, responds very quickly and makes any changes instantly. Instead, we conducted various tests of the bot.
I recommend.

Vlad Y. | Safe

| Response review

11 January 27 USD
Bot script for sending messages

Quality

Professionalism

Cost

Contactability

Deadlines

Completed quickly, efficiently, everything as needed and according to the specifications)
I recommend the performer!

Denis Shkarpenko | Safe

| Response review

5 January 45 USD
Telegram bot

Quality

Professionalism

Cost

Contactability

Deadlines

You completed the work quickly and very efficiently.

Vlad Y. | Safe

| Response review

8 December 2025 111 USD
Create a Telegram chat bot with automation for application approval (Excel document with macros).

Quality

Professionalism

Cost

Contactability

Deadlines

Everything was done professionally, within the agreed deadlines. I recommend the specialist for creating bots.

Nadiia Winowicz | Safe

| Response review

22 November 2025 40 USD
Bot for sending messages in Telegram

Quality

Professionalism

Cost

Contactability

Deadlines

I recommend, quickly and efficiently!

Serhii M. | Personal | Response review

16 November 2025 16 USD
Parser for routing

Quality

Professionalism

Cost

Contactability

Deadlines

Everything was done very quickly and efficiently. It is pleasant to deal with a person who truly understands what they are doing. Georgiy works clearly, confidently, and without unnecessary questions; the result is exactly what was needed.

Everything was done without delays, neatly and professionally. The performer can be trusted with complex tasks; the work is truly of high quality.

I will definitely reach out again. I definitely recommend. The result exceeded my expectations!!!

Hanna Mykhniuk | Safe

| Response review

Activity

Latest proposals 10	Budget	Added	Deadlines	Proposal
TG BOT for Auto Ria 156 USD	156 USD	25 Jun	3 days
AI Resume Analyzer 602 USD	602 USD	24 Jun	7 days	rejected
Development of a Telegram store for a clothing brand. Product catalog, sizes, cart, order processing. 178 USD	178 USD	23 Jun	4 days
Bot for receiving/searching applications 28 USD	28 USD	22 Jun	2 days
Business card website for Auto Service 178 USD	178 USD	21 Jun	2 days
Turnkey Telegram bot: daily reminders and rotation of duty personnel with monitoring 66 USD	66 USD	20 Jun	1 day
Telegram bot for auto service 39 USD	39 USD	20 Jun	1 day
Telegram bot for hiring/searching employees. For job searching. 25 USD	25 USD	18 Jun	1 day
Setting up a monthly subscription for a Telegram channel/group 36 USD	36 USD	17 Jun	1 day	rejected
A Telegram bot is needed. The programming language is Python, and the bot must have an admin panel for processing requests. 45 USD	45 USD	17 Jun	2 days	rejected

Heorhii Zhukov

Rating

CV

Skills and abilities

Programming

Portfolio

Reviews and compliments on completed projects 24

17 June 181 USD Website development

12 June 446 USD Website development

10 June 22 USD Improvement of the Telegram bot

30 May 40 USD Improve the bot

22 May 89 USD Improve the bot

11 May 100 USD Telegram bot with Claude API and Google Drive — personal AI agent

8 May 35 USD It is necessary to set up Windows VDS automatically for operation through Windows App / Remote Desktop.

6 May 94 USD Development of a Telegram bot for a garbage collection service

6 May 56 USD Auto Proxy Setup for Windows VDS (SOCKS5)

2 May 27 USD Development of a Python script for local export of Telegram channels (open/closed)

10 February 29 USD Improvement of the bot

4 February 29 USD Data parsing from the website

30 January 42 USD Proposal parser from the website

29 January 67 USD Group in Telegram using a bot for solving some tasks

22 January 67 USD Regular data collection from Rozetka by categories (with reviews and photos)

20 January 38 USD Improvement of the Telegram bot (training)

18 January 22 USD Telegram bot for parsing channels

14 January 29 USD Set up automatic publication of messages about the start and end of air alerts in the Telegram channel.

12 January 45 USD Development of a Telegram bot with AI for communication with candidates and selection of applications.

11 January 27 USD Bot script for sending messages

5 January 45 USD Telegram bot

8 December 2025 111 USD Create a Telegram chat bot with automation for application approval (Excel document with macros).

22 November 2025 40 USD Bot for sending messages in Telegram

16 November 2025 16 USD Parser for routing

Activity

17 June 181 USD
Website development

12 June 446 USD
Website development

10 June 22 USD
Improvement of the Telegram bot

30 May 40 USD
Improve the bot

22 May 89 USD
Improve the bot

11 May 100 USD
Telegram bot with Claude API and Google Drive — personal AI agent

8 May 35 USD
It is necessary to set up Windows VDS automatically for operation through Windows App / Remote Desktop.

6 May 94 USD
Development of a Telegram bot for a garbage collection service

6 May 56 USD
Auto Proxy Setup for Windows VDS (SOCKS5)

2 May 27 USD
Development of a Python script for local export of Telegram channels (open/closed)

10 February 29 USD
Improvement of the bot

4 February 29 USD
Data parsing from the website

30 January 42 USD
Proposal parser from the website

29 January 67 USD
Group in Telegram using a bot for solving some tasks

22 January 67 USD
Regular data collection from Rozetka by categories (with reviews and photos)

20 January 38 USD
Improvement of the Telegram bot (training)

18 January 22 USD
Telegram bot for parsing channels

14 January 29 USD
Set up automatic publication of messages about the start and end of air alerts in the Telegram channel.

12 January 45 USD
Development of a Telegram bot with AI for communication with candidates and selection of applications.

11 January 27 USD
Bot script for sending messages

5 January 45 USD
Telegram bot

8 December 2025 111 USD
Create a Telegram chat bot with automation for application approval (Excel document with macros).

22 November 2025 40 USD
Bot for sending messages in Telegram

16 November 2025 16 USD
Parser for routing