Kateryna Kolupai
Offer Kateryna work on your next project.
Private message
Offer work
Zaporizhzhia,
Ukraine
4 days 3 hours back
available for hire
2 Safes completed
3 months 19 days back
1 client
2 proposals made
age 27 years
on the service 6 years
Rating
Successful projects
Average rating
Skills and abilities
Design & art
Services
Portfolio
-
AI photo editing through text commands
AI & Machine Learning
I created a system that allows editing photos with simple text commands. The user can change details in the photo, replace the background, or completely change the style — without graphic editors.
How it works:
the user sends a photo and describes what needs to be changed;
… the system understands the request in any language and correctly processes the edits;
it supports dialogue — changes can be gradually clarified;
the result is immediately shown in a convenient format.
Result for the user:
quick photo editing without a designer;
preservation of the composition and quality of the original;
the ability to create many variations of one image;
a ready foundation for e-commerce and "virtual try-on" solutions.
Tools: n8n, Flux 2 Pro, Claude, JavaScript, API integration.
-
AI screenwriter for short videos (Shorts / Reels / TikTok)
AI & Machine Learning
I created an AI system that automatically prepares scripts for short videos. The user sends an idea or text — the system transforms it into a ready structure with 5 scenes for video production.
How it works:
AI analyzes the input text and logically breaks it down into scenes (beginning, development, conclusion);
… it fixes the character and style so they remain consistent across all scenes;
for each scene, it generates:
a professional AI prompt,
instructions for animation,
text for the narrator;
it supports edits through dialogue (for example, changing the style or atmosphere).
Result for the user:
quick script preparation without manual work;
stable structure and consistency of characters;
savings of up to 80% time on pre-production;
a ready base for video generation in AI tools.
Tools: n8n, Claude 4.5 Sonnet, context memory, advanced prompt engineering.
-
AI image generation system through dialogue
AI & Machine Learning
I created an intelligent system that allows generating and improving images in simple terms. The user does not need to know prompt engineering — the system itself transforms the idea into an accurate technical request and produces ready-made images.
How it works:
the user describes the idea in any language;
… the AI agent translates and optimizes the request for the generation model;
remembers previous edits (style, light, mood) and takes them into account;
generates 3 vertical images (1080×1920) per request;
shows the result immediately in the chat without unnecessary links.
Result for the user:
quick creation of visual content without technical knowledge;
convenient refinement through dialogue;
stable quality for Stories, Shorts, and advertising;
low cost of generation.
Tools: n8n, Claude 4.5 Sonnet, Flux 2 Dev (ModelsLab), JavaScript.
-
AI Video Factory — automatic creation of Shorts / TikTok
AI & Machine Learning
I have developed a system that automatically generates short videos (Shorts, TikTok, Reels) based on the user's idea. Instead of manual editing, the AI itself creates smooth animations between key frames and produces a finished clip in seconds.
How it works:
the user provides an idea and key frames (beginning / end);
… the AI agent transforms the idea into a technical prompt for the video model;
the system generates video with cinematic movement;
the result can be refined through dialogue (“add dynamics,” “make it smoother”).
Result for the user:
quick video creation without editing;
stable quality for commercial content;
minimal production costs;
perfect for mass testing of creatives and advertising.
Tools: n8n, Anthropic Claude 4.5 Sonnet, Video Generation API (Seedance), JavaScript.
-
AI agent editor and content critic
I created an AI agent that automatically checks and edits scripts for short videos. It works like a strict editor: it evaluates the text according to clear rules and does not allow weak content into production.
How it works:
the agent analyzes the script and assigns a score on a 100-point scale;
… it checks volume, style, and prohibited patterns (clichés, unnecessary words, etc.);
if the score is below the threshold, the script is automatically rewritten;
the output is clean text and a final score without "extra" AI comments.
Result for the user:
stable quality of scripts without manual proofreading;
content better retains attention;
video production can be scaled even for a single creator;
a ready solution for integration into bots or frontend.
Tools: n8n, Anthropic Sonnet 4.5, LangChain, logic checks, and session memory.
-
Multi-Agent system for video scenario generation
AI & Machine Learning
I created an automated pipeline in n8n for generating quality scripts for short videos (YouTube Shorts / TikTok). The system operates through several AI agents and controls the quality of the content itself.
How it works:
The Author agent generates a short script (55–62 words) according to clear rules;
… The Critic agent evaluates the text on a 100-point scale and initiates a revision if the quality is below the threshold;
The system separates the finished script from technical evaluations;
The final result is automatically sent to Airtable for further production.
Result for the user:
Stable quality of scripts without "watery" AI text;
Texts adapted to Shorts algorithms;
Minimum manual edits;
Ready content for large-scale video production.
Tools: n8n, Claude Sonnet, LangChain, JavaScript, Airtable.
-
AI analytics of short videos (Shorts / Reels)
AI & Machine Learning
I have developed an automated system that analyzes short videos (YouTube Shorts, Reels, TikTok) and provides quick, clear feedback for improving reach and attention retention.
How it works:
the user submits a video or link;
… AI analyzes the first seconds, editing, pace, lighting, sound, and 9:16 format;
the system generates clear recommendations: what to improve, what hinders virality;
the result is delivered almost instantly via webhook.
Result for the user:
professional video audit in a few seconds;
understanding why the video does not "hook";
time savings on viewing and edits;
consistent content quality for bloggers and agencies.
Tools: n8n, Google Gemini 2.5 Flash, JavaScript, Webhooks.
-
Automatic tracker of AI-token expenses (n8n)
AI & Machine Learning
I created an automation that tracks the usage of AI models (GPT-4, GPT-3.5, Claude) and shows where the AI budget is actually being spent. The system works as a separate "monitoring layer" and can connect to any n8n workflows.
How it works:
automatically collects data on AI model calls in the workflow;
… counts prompt and completion tokens for each model;
groups expenses by GPT-4, GPT-3.5, Claude;
stores all statistics in Google Sheets in an understandable format.
Result for the business:
complete transparency of AI expenses;
clear understanding of which models and prompts are "eating" the budget;
no manual checks and surprises in expenses;
easier to control and scale AI systems.
Tools: n8n, n8n API, Google Sheets, OpenAI, Anthropic.
-
AI system for automatic article creation
AI & Machine Learning
I created an automation that transforms a short request or idea into a ready-made blog article. The system prepares the technical task itself, gathers relevant information from the internet, and generates a complete, structured text that is immediately published in Notion.
How it works:
the user provides a brief description of the topic;
… AI formulates a clear technical task for the article;
automatically collects facts and context from the web;
generates an SEO-optimized article with headings and lists;
publishes the finished material in Notion.
Result for the user:
quick creation of quality content without manual work;
articles with a logical structure and verified facts;
savings of up to ~70% time on preparing materials;
a convenient process from idea to publication.
Tools: n8n, OpenAI (GPT-4.1), Perplexity API, Notion.
-
AI system for task creation in Notion from messages
AI & Machine Learning
I created an automation that transforms regular text messages into ready tasks in Notion. The user writes a message in free form — the system analyzes the text, generates a title and description of the task in Ukrainian, and adds it to the appropriate Notion database.
How it works:
the message is received via webhook;
… AI analyzes the content, determines the essence of the task and the responsible person;
generates a short clear title and a detailed description;
automatically creates a task in the corresponding Notion database.
Result for the user:
there is no need to manually format tasks;
no important detail from the messages is lost;
all tasks have a unified, neat format;
convenient organization of work for the team.
Tools: n8n, OpenAI, Notion, Webhook.
-
Automatic monitoring of marketing projects
AI & Machine Learning
I created an automation that checks for new marketing projects every 5 minutes across several freelance platforms and sends only relevant ones to Telegram and Google Sheets.
Platforms:
Freelancer.com (API), Freelance.ua, Weblancer, PeoplePerHour (RSS)
…
How it works:
the script runs on a timer;
collects new projects from API and RSS;
filters job postings by keywords marketing;
checks for duplicates;
saves all projects in Google Sheets;
sends new tasks to Telegram in a convenient format.
Result for the user:
all marketing projects from different exchanges in one place;
instant notifications without spam and duplicates;
time-saving and fully automated process.
Tools: n8n, API, RSS, Google Sheets, Telegram.
-
Automatic monitoring of projects from Upwork (n8n)
AI & Machine Learning
I created an automation that automatically tracks new projects on Upwork, saves them in Google Sheets, and immediately sends notifications in Telegram.
How it works:
the script runs on a timer;
… it retrieves new job postings from Upwork via API;
it checks if they have been posted before;
it adds new projects to Google Sheets;
it sends a convenient notification in Telegram.
Result for the user:
no need to manually update Upwork;
all job postings are saved in the spreadsheet;
instant notifications about new projects;
time savings and a completely automated process.
Tools: n8n, Upwork API, Google Sheets, Telegram.
Reviews and compliments on completed projects 2
13 February
29 USD
Adding products
Quality
Professionalism
Cost
Contactability
Deadlines
The work was completed on time. I recommend.
Oleksii Vinichenko
| Personal | Response review
29 January
16 USD
Adding products
Quality
Professionalism
Cost
Contactability
Deadlines
The work was done quickly and efficiently. I recommend.
Oleksii Vinichenko
| Personal | Response review
Activity
| Latest proposals 10 | Budget | Added | Deadlines | Proposal | |
|---|---|---|---|---|---|
|
AI AGENT
|
|||||
|
Open cart 4. Journal3. Filling the online store
29 USD
|
|||||
|
Content filling - populating website cards on WordPress
113 USD
|
|||||
|
Filling an online furniture store
23 USD
|
|||||
|
Content Manager / Product Filling for the Online Store DAILYOPT
90 USD
|
|||||
|
It is necessary to list 100 product items on the Rozetka marketplace.
90 USD
|
|||||
|
Filling promo 50 cards
34 USD
|
|||||
|
Content manager for an online medical-themed store
23 USD
|
|||||
|
Transfer goods to the marketplace Rozetka
135 USD
|
|||||
|
Administrator for managing the site on Khoroshop
16 USD
|
