Wiktoria Jarema
Offer Wiktoria work on your next project.
Private message
Offer work
Krakow,
Poland
4 months 12 days back
available for hire
on the service 10 months 6 days
Rating
Successful projects
Average rating
Skills and abilities
Portfolio
-
Comprehensive Recruitment Process Automation
Project scope:
The project involved implementing an automated recruitment system along with omnichannel communication, an AI chatbot, and document processing. The implementation was divided into 9 stages. Each was delivered on time with documentation and testing.
Stage 1: Database design + automations (Airtable + Make)
… Development of candidate and client database structure in Airtable
Integration of forms (WordPress/Typeform) with Make
Automatic filtering by location, language, experience
Sending reminders about missing documents (24h)
Generating candidate cards in PDF + sending to client
Registering candidate statuses in the system
Stage 2: Integration of communication channels (Omnichannel)
WhatsApp Business API (Twilio / 360Dialog)
Communication via email, SMSAPI, Messenger, Viber (MessageBird)
Testing and optimization of each channel
Stage 3: Multilingual support (PL / EN / UA)
Message templates in 3 languages
Automatic language assignment based on data
Stage 4: Client form + contract generation
Client form with company data
Generating contracts and attachments in PDF based on data
Integration with Dropbox Sign (e-signature)
Stage 5: Automatic offer sending
Checkbox "I want an offer" in the form
Generating personalized PDF offers
Follow-up email after 3 days of no response
Stage 6: AI Agent – recruitment chatbot
Chatbot on website, WhatsApp, Messenger
Conversation scenarios in PL/EN/UA
Reminders about missing documents
Weekly PDF report summarizing data
Stage 7: Recruitment team supervision system
Dashboard in Airtable (view: candidates by recruiter, no activity for 48h, delays)
Daily and weekly reports (email, Slack)
Scoring + alerts about inactivity
Integration with AI – performance summaries (ChatGPT)
Stage 8: Cost invoice handling (OCR + database)
OCR of invoices from PDF (Google Vision API)
Automatic data reading and saving to database (Airtable / Sheets)
Statuses: paid / unpaid
Payment deadline reminders
Stage 9: Voicebot – automatic callback
Implementation of voicebot (Twilio / Voximplant)
Automatic callback after failed call
SMS with link to contact form
Conversation scenarios in PL/EN/UA
-
Automation of product instruction translation
Project goal:
The aim of the project was to create an automated process for translating product instructions into multiple languages, without the need for manual copying and pasting of content. The client expected a solution that allows for flexible, fast translation of text files (including DOCX, ODS) with the possibility of further expansion.
Scope of work:
… Stage 1 – Design and configuration of workflow:
Create an automation workflow in the n8n environment (with an alternative option for implementation in Make.com / Integromat).
Integration with translation tools – depending on the version:
Google Translate API (fast, widely supported option)
DeepL API (better translation quality for selected languages)
Stage 2 – Handling input files:
Automatic detection of new text documents placed in a dedicated folder on Google Drive or uploaded via form/upload.
Recognition of DOCX and ODS document structures and text extraction.
Stage 3 – Translation and saving results:
Sending the extracted text to the selected translation API.
Handling multiple languages (including dynamic selection of the target language).
Preserving document formatting structure (e.g., headers, lists, paragraphs).
Generating a new document in the output format (DOCX / ODS) and saving it in a separate folder or target system.
Option to add a draft version for later editing.
Stage 4 – Scalability and flexibility:
Workflow designed to be scalable: easy addition of new languages, file sources, and additional steps (e.g., OCR, transliteration, correction).
Error handling, event logging, and automatic alerts in case of failed translation or format errors.
Final result:
A fully automated system for translating product instructions, which enables:
Automatic detection of new files with content to be translated.
Support for translations into multiple languages while maintaining the original format.
Integration with Google Drive for storing completed documents.
Reduction of time and costs associated with manual content translation.
Readiness for further expansion – e.g., review of translated texts, AI correction, or distribution to clients.
-
PHP/Symfony Library or n8n automation for processing li
Project goal:
The aim was to create a scalable solution enabling automatic processing of approximately 3000 links daily from the Allegro platform, extracting key product information. The client allowed implementation in two variants: as a PHP library for the Symfony framework or as an automation process based on the n8n platform.
Scope of implementation:
… Variant 1 – PHP/Symfony Library:
Create a dedicated library handling the processing of single and bulk Allegro links
Parsing data from HTML and dynamic elements (JS-rendering) using a headless browser (e.g., Puppeteer)
Output structure in JSON with the following fields:
Offer name
Offer ID
Seller name
Number of recent purchases
Links to images (main, thumbnails, gallery – if available)
Reviews and ratings (count, average)
Information about available offer variants
Display status: sponsored / promoted / regular
Delivery details (e.g., "delivery tomorrow", cutoff times, SMART options)
Special markings: SMART, SUPERCENA, Smart! Weeks
Technical measures to bypass Allegro security:
IP rotation
Proxy
Headless browser
Request limiting
Randomization of User-Agents
Variant 2 – Automation using n8n:
Create a workflow for bulk processing of links and retrieving product data
Use of nodes such as HTTP Request, HTML Extract, Custom JavaScript, and external connectors
Configuration of request limiting system and proxy support
Ability to export data to JSON, Google Sheets, or a database
Technical difficulties and solutions:
Allegro uses dynamic data loading, which required implementing a page rendering mechanism in a browser without interface (headless browser).
To maintain stable operation under high query volume, a system of IP address rotation and automatic proxy session restart was implemented.
An error logging and retry system was implemented to handle limit breaches or server errors.
Final result:
Two independent solutions were developed:
A Symfony library that can be invoked from CLI or API and returns data in a unified JSON format.
An n8n workflow enabling bulk link processing, data saving, and further processing.
Activity
| Latest proposals 10 | Budget | Added | Deadlines | Proposal | |
|---|---|---|---|---|---|
|
Check the ready ManyChat bot and connect the calendar + Ads Manager + CRM
290 USD
|
|||||
|
Development: AI assistant for leads and clients
823 USD
|
|||||
|
Facebook marketplace automation
960 USD
|
|||||
|
Bundle for auto-creation and posting of content in social networks.
349 USD
|
|||||
|
Working with API connection for AI assistants
500 USD
|
|||||
|
Business automation
800 USD
|
|||||
|
Assistant program (bot) in the Facebook environment
350 USD
|
|||||
|
Development of a Telegram userbot with personality and memory (based on Chat GPT)
650 USD
|
|||||
|
Full automation of business processes in the company
1162 USD
|
|||||
|
Creating a Telegram bot for a master class with payment
232 USD
|
