Illia S.
Private message 
Offer work 
Kyiv,
Ukraine
available for hire
Rating
Language proficiency level
Українська: fluent
Русский: fluent
English: advanced
Skills and abilities
Programming
-
AI & Machine Learning
from 564 USD for project
-
Bot Development
from 226 USD for project
-
Cryptocurrency & Blockchain
from 90 USD for day
-
Data Parsing
from 113 USD for project
-
Databases & SQL
from 45 USD for day
-
Desktop Apps
from 226 USD for project
-
Javascript and Typescript
from 338 USD for project
-
Python
from 90 USD for day
-
Web Programming
from 45 USD for day
Design & art
-
AI Art
from 90 USD for variant
-
Web Design
from 226 USD for project
Services
-
AI Consulting
from 45 USD for hour
-
Data Processing
from 45 USD for project
-
Online Stores & E-commerce
from 226 USD for project
-
Payment Systems Integration
from 45 USD for day
-
Website Development
from 226 USD for project
Photo, Audio & Video
-
AI Speech & Audio Generation
from 113 USD for variant
-
Video Creation by Artificial Intelligence
from 113 USD for variant
Outsourcing & consulting
-
Project Management
from 902 USD for month
Portfolio
-
564 USD Real-Time Voice Translator -
AI & Machine Learning
A universal voice translator for Windows that works with any application - Zoom, Discord, games, browsers. Not a plugin or extension, but an OS-level solution with under 300 ms end-to-end latency.
Two parallel pipelines provide bidirectional translation: your speech is translated and routed to a virtual microphone for the other person, while their audio is captured from system output, translated, and played back through your speakers. Any language pair is supported.
… Built with streaming speech recognition, LLM-powered translation with caching, neural text-to-speech, voice activity detection, and low-level Windows audio integration. 1500+ lines of production code.
Stack: Python, Streaming STT, LLM Translation, Neural TTS, VAD, WASAPI, Virtual Audio Driver
-
564 USD Multi-Agent AI System on OpenClaw
AI & Machine Learning
AI agent swarm on the OpenClaw platform with a central coordinator and specialized agents. Each agent runs as an isolated session with its own workspace, long-term memory (vector embeddings), toolset, and Telegram bot.
Architecture: Main (coordinator, Opus) manages Sonnet-powered agents — Social (content planning, analytics, outreach, trend monitoring), Mail (Gmail API via OAuth 2.0 — reading, sending, classification), and a startup platform monitoring agent.
… Cron jobs ensure continuous operation: email checks every 2h, morning briefings, social analytics, content generation. Models are matched to task complexity — Opus for coordination, Sonnet for main agents, Haiku for routine crons.
Inter-agent communication: fire-and-forget via sessions_send (solving queue serialization — one run per session). Agents report to the coordinator + duplicate to the owner via Telegram. Browser automation via Playwright: authentication, scraping, form filling.
Stack: OpenClaw, Node.js, Python, Playwright, Gmail API (OAuth 2.0), Telegram Bot API, Anthropic Claude (Opus/Sonnet/Haiku), BGE-M3 embeddings.
-
338 USD Autonomous Browser Agent — 30/30 in 51s
Web Programming
Autonomous browser agent that solves all 30 steps of the Browser Navigation Challenge in 51 seconds. Zero errors, zero LLM tokens, $0 runtime cost.
Uses React Fiber tree traversal — locates the challenge component's onComplete callback in the React internal tree and invokes it directly with a valid proof object. One universal method handles all 26 challenge types (hidden_dom, drag_drop, canvas, websocket, shadow_dom, etc.).
… XOR session decoding from sessionStorage to obtain all 30 verification codes upfront. Automatic obstacle dismissal: cookie banners, popups, overlays, fake buttons (15-25 per step).
Stack: Node.js + Playwright (Chromium). Single file ~500 lines. Video recording + JSON report with per-step timing.
-
338 USD N8N Workflows Parser с Semantic Search
AI & Machine Learning
Parser of open free n8n templates with vector embeddings generation for semantic search. The system extracts titles, descriptions, tags, workflow authors, and generates 384-dimensional vectors using sentence-transformers. The search works semantically rather than by keyword matching — it finds relevant results even when the query is phrased differently.
Desktop GUI on PyQt: parser settings (pages count, rate limiting, user-agent), embedding generation, results table with relevance score sorting. Error handling for network timeouts and CloudFlare challenges. Export to JSON/CSV for database import.
… Raw data in downloads/, processed results in results/, embeddings in binary format for fast download. Custom similarity search on cosine distance with a threshold of 0.7.
The result is semantic search in the n8n workflow database instead of manual browsing + the ability to use, copy, and work with these n8n templates.
In the future, we plan to teach AI to read and learn the necessary templates for automation automation.
-
226 USD Markdown - Word/PDF Converter
Python
Production-ready solution for converting between Markdown and Word/PDF in both directions. Markdown → Word/PDF preserves LaTeX formulas (inline/display), code blocks, tables. Word/PDF → Markdown converts OMML formulas to LaTeX, tables to pipe format, extracts images. GitHub Flavored Markdown support out of the box.
Desktop GUI on PyQt with drag & drop, two tabs for each direction, JSON formatting schemes for batch processing. Portable Pandoc included—zero dependencies setup. Libraries: mammoth for .docx parsing, pypandoc for conversion, custom regex for formulas.
… Handles complex cases: nested lists, tables in tables, mixed formulas. Settings persistence for recurring tasks.
-
338 USD Trend Image Generator on Multi-Agent Architecture
AI & Machine Learning
A full-cycle image generation automation system — from market analysis to finished content with metadata. Seven specialized agents work sequentially: DailyFocusAgent analyzes current events through the Perplexity API, TrendAgent identifies commercial gaps in a three-stage research process, and PhotoAgent generates options on Nano Banana Pro with AI recommendations for styles (safe/risky) and formats (square/landscape/portrait).
The key advantage is automatic QA through VisionAnalysisAgent: each image is evaluated according to 10 criteria (composition, lighting, artifacts, commercial viability), and if the score is low, a correction loop is launched with regeneration based on specific feedback. The result is top-quality images at a total cost of all processes less than one premium generation.
… QualityAgent checks technical parameters, UploadAgent prepares titles/keywords/categories, AnalyticsAgent logs costs and attempts in JSON. GUI on Tkinter for manual control and dashboard. Suitable for any type of image — from commercial photography to illustrations and digital art.
The result: autonomous generation of commercially viable content with built-in quality control and complete analytics.
-
451 USD TA crypto module
Cryptocurrency & Blockchain
Traders spend hours looking at charts, searching for patterns manually. I have automated what usually requires years of experience.
Real-time technical pattern recognition system for cryptocurrencies: 64 patterns (candlestick, graphic, harmonic) plus 80+ indicators on hourly candles with Binance Futures. Modular architecture: TA-Lib detects candlestick patterns (Hammer, Doji, Engulfing, Morning/Evening Star), custom algorithms recognize graphical patterns (Head & Shoulders, Double Top/Bottom, Triangles, Flags, Wedges), Fibonacci calculations find harmonic patterns (Gartley, Bat, Butterfly, Crab, Shark, ABCD).
… React frontend with Lightweight Charts renders the dark Binance theme — green and red candlesticks, as traders are accustomed to. But the main thing is not the visuals, but the forecast: the system outputs specific target price levels with probabilities based on the weighted accuracy of each pattern. Type weights take into account that harmonic patterns are more reliable than candlestick patterns, and the confluence of several patterns gives the strongest signals.
Only recent data (the last 30-100 candles) is analyzed — no outdated patterns from last month. Dynamic rounding: for BTC, it shows decimal places, for SHIB — whole numbers. The consensus panel summarizes all patterns and gives an overall verdict: LONG, SHORT, or NEUTRAL.
-
226 USD AI technical drawing generator with Nano Banana pro
AI & Machine Learning
When conventional AI image generators cannot meet technical requirements, I created a solution that understands GOST and speaks the language of engineers.
This is a full-fledged image generation system on Google Gemini 3 Pro, the basis of thinking for Nano Banana Pro, which goes far beyond simple pictures. Multi-generation runs 4 variants in parallel via ThreadPoolExecutor, each with a unique variation seed for diverse results. Reference images (up to 14) are passed as a style guide — AI analyzes them and applies the visual style to the new generation.
… But the main feature is full integration with the CAD world: automatic export to DXF via ezdxf (vector contours for AutoCAD), generation of A1 format PDF (841×594 mm) with the correct DPI for printing, and stamping according to DSTU (Ukrainian GOST 2.104). The stamp is generated by software: a frame with 20 mm indents on the left and 5 mm on the other sides, 185×55 mm cells with fields for Development, Verification, Approval, scale, mass, and organization. The text is automatically transferred to 2 lines or reduced if it does not fit.
Result: engineers receive not just pictures, but ready-made drawings with the correct stamps and in the required formats.
Translated with DeepL.com (free version)
-
79 USD Automation of processes using n8n
AI & Machine Learning
A bot with internet access that has tools - a workflow search engine and a doctor. The bot accepts documents, voice messages, and text. It can analyze the user's medical data and remember it.
The doctor's tool has its own memory and analytics for each individual patient and can also access the internet using the internet search tool.
… The internet search tool has several operating modes depending on the user's request.
The entire workflow is managed by the BOT, which can also remember the chat and adapt to each user.
-
23 USD Code review and optimization
Python
I will conduct a professional code review focusing on performance, clean architecture, and best security practices. My goal is to optimize your code to ensure maintainability, efficiency, and scalability. You will receive detailed recommendations and practical solutions to improve code quality and its long-term sustainability.
Reviews and compliments on completed projects 6 1
17 April
298 USD
SEO processing of products,
Everything is great, I will continue to reach out.
16 March
439 USD
AI Agent
Everything is great, I am reaching out for the first time!
14 March
439 USD
AI Agent
Quickly, everything is great! I recommend.
The performer is diligent, always in touch. But we will still be finishing the project.
25 September 2025
169 USD
Selection of LLM for reading emails with JSON output + Python libraries
The project was not completed, the contractor tried, but it didn't work out.
19 September 2025
271 USD
Data parser and API service modernization
The developer completed a full project ahead of schedule for creating a data parser based on Express.js and JavaScript, demonstrating a high level of technical preparation and responsibility. During the work process, he showed a systematic approach: starting from analyzing requirements and designing the architecture of the service, to testing and documenting key decisions.
The code is written with a clear emphasis on structure and readability. The logic for handling requests, parsing, and further data processing is clearly separated, which significantly simplifies the maintenance and development of the project. It is important to note that the developer paid attention to error handling and logging — this allows not only to track the current state of the service's operation but also to respond promptly to non-standard situations.
His approach to scalability deserves special mention. The project was designed so that it could be easily adapted to new data sources or integrated with other services. The use of Express.js turned out to be a successful solution: it provided ease of API configuration and flexibility for future enhancements.
In addition to the technical side, I want to highlight the communication. The developer was always open to discussing details, made suggestions for improvements, and argued convincingly for the solutions he considered most effective. He demonstrated the ability to work with priorities: focusing on the main tasks without getting bogged down in secondary ones, while still maintaining the quality of implementation.
In the end, a reliable tool was created, which became an important part of the project. The work was done qualitatively and on time, and the code can be rightly called a model of competent development in JavaScript. I am confident that this experience demonstrates the high professional level of the developer and his readiness for more complex and large-scale tasks.
3 August 2025
23 USD
Retrieving data from the website
The project was completed on time, taking into account additional nuances that arose during the work process. I recommend collaboration
Activity
| Latest proposals 10 | Budget | Added | Deadlines | Proposal | |
|---|---|---|---|---|---|
|
Freelance project
609 USD
|
|||||
|
Freelance project
250 USD
|
|||||
|
Freelance project
451 USD
|
|||||
|
Freelance project
34 USD
|
|||||
|
Freelance project
1645 USD
|
|||||
|
Freelance project
2000 USD
|
|||||
|
Freelance project
13 710 USD
|
|||||
|
Freelance project
274 USD
|
|||||
|
Freelance project
250 USD
|
|||||
|
Freelance project
564 USD
|