Three high‑impact builds showcasing AI automation and modern web engineering.
Client ProjectPythonGPT-4MagentoNode.js
Advanced AI Assistant Chat Bot
Designed and deployed an AI-powered quoting engine and flue advisor across 12 e-commerce stores in Europe. The system auto-generates accurate quotes, understands technical requirements, and integrates with Magento, saving over €100k/year in staffing.
Client ProjectPythonSeleniumConcurrency
Automated Price Scraper
Built a scalable, multi-country scraper to extract live prices from e-commerce sites across Europe. It uses a persistent browser pool with advanced handling for popups, JavaScript content, and multilingual cookie banners.
ReactAPI IntegrationAxios
Interactive Weather Forecast
A React-powered weather interface that delivers real-time and upcoming conditions for any location. It uses structured state management and API integrations to present clear, dynamic forecasts with smooth updates. The design focuses on responsiveness and ease of use, offering a polished way to explore weather data through an intuitive, modern UI.
I designed and built a full AI-powered assistant used across 12 European e-commerce shops.
The system automates technical support, designs flue systems, recommends stoves, generates quotes,
emails PDFs, and basically does the work of an entire support team, without ever sleeping.
Technologies
GPT-powered reasoning system
Node.js backend + JSON → PDF quote builder
Magento REST/SOAP integration
Vision processing for photo/sketch interpretation
Supabase conversation database
Netlify-hosted widget for instant deployment everywhere
Key Features
AI-guided Flue System Wizard with technical reasoning
Automatic quote generation (JSON → PDF)
Magento integration with SKU mapping and email delivery
Vision input (photo/sketch) to detect roof pitch, wall exits, or flue layout
Conversation history database + source-linked knowledge base
Works across every country site without additional staff
Key Challenges & Solutions
Highly technical reasoning needed: Built multi-step guided wizards that mimic real installer logic (diameter, offsets, roof pitch, regulations).
Slow, manual quoting process: Developed a backend that converts user inputs directly into complete product lists, PDFs, and Magento-compatible JSON.
Multiple regional shops with different languages: Created a single AI system deployable across 12 sites, translation-ready and auto-updating via Netlify.
Users sending photos and sketches: Added a vision pipeline to auto-detect features and pre-fill wizard inputs.
Business Impact
Handles 120–360 quotes weekly across all regions
Saves 40–120 staff hours per week → €50,000–€125,000 yearly
Increases conversions by 10–20% thanks to instant expert-level answers
Pays for itself in 1–3 months
Creates a scalable, multilingual AI infrastructure for future expansion
24/7 availability with instant responses
Outcome
A robust AI assistant that automates technical support and quoting across multiple e-commerce sites.
Significant cost savings and efficiency improvements.
Happier customers thanks to instant expert-level help.
Reduced mistakes and improved installation accuracy.
More revenue through smarter, faster guidance and quoting.
A future-proof system that can be easily updated or expanded.
This project transformed a complex, highly technical support workflow into a smooth, automated, and multilingual AI-driven experience. Running 24/7 with zero additional staffing.
Custom AI App Case Study
A full-stack React environment for deploying lightweight AI tools with live interaction.
E-Commerce Price Scraper for European Markets
(Python • Selenium • Multi-Country Automation)
I developed a high-performance automated price scraper to crawl e-commerce sites across nine European countries, handling dynamic content, multilingual cookie banners, and anti-scraping measures.
Technologies
Python, Selenium (Firefox/GeckoDriver), Pandas
Threading + Queue for concurrency
Multi-sheet Excel export (one sheet per country)
Key Features
Scrapes hundreds of URLs across multilingual markets
Persistent browser pool architecture (massive speed boost)
Intelligent popup/cookie banner dismissal
95%+ success rate across all websites
Automated Excel report generation
Key Challenges & Solutions
Browser crashes: Migrated from Chrome to Firefox; built a full cleanup process for stability
Slow runtime: Rebuilt core logic using a producer–consumer model → 80% faster overall
Dynamic content blocking: Added a flexible multilingual popup handler using keyword/selector heuristics
Outcome
A stable, fast, and scalable scraper that reliably collects international pricing data and outputs it into a structured, ready-to-use Excel report.
