Automation of data collection: web scraping, PDF parsing and desktop applications
About the service
I develop turnkey automation and data collection tools in Python. Full cycle: from the analysis of your task to the working script, which can be run without my participation.
The service includes:
Analysis of the data source (website, desktop program, PDF, Excel) and selection of the optimal tool.
Writing code with error processing, logging and a checkpoint system (if the script falls in the middle, it will continue from the stop, rather than starting over).
Save the result in a convenient format: Excel, JSON, CSV, PDF, LaTeX or direct upload to your CRM / database.
Launch instructions and brief documentation.
7 days of free edits after delivery.
What I do most often:
Searching sites with antibot protection (Cloudflare, Captcha, Dynamic JS).
Automation of desktop programs that do not have an API (1C, PROSIS, closed vendor clients).
Extracting data from “curved” PDF documents: tables without borders, scans, multi-page manuals.
Building ETL-piplins: collection → cleaning → normalization → unloading into the report.
Experience:
3 years of commercial development. During this time, he made 20+ working tools for one large customer - a distributor of special equipment. Scripts collect spare parts catalogs and service documentation from 6 global vendors: Caterpillar, Volvo, Komatsu, JCB, CNH Industrial (Case/New Holland/Steyr).
In parallel, he developed parsers of equipment marketplaces (TractorHouse, MachineryTrader) - collecting millions of lots with parallel downloading of images and uploading to Excel.
All instruments have been in production for years, not once shown and forgotten.
What makes me different from other artists:
I make scripts that don't fall. My projects include multi-level checkpoints, retry logic and file validation. If the Internet falls off, the server sits down or the site changes the layout - the script will recover correctly, and not lose 8 hours of work.
I accept what others refuse. GUI without API, PDF without structure, sites with rigid antibot protection, legacy applications 10 years ago. Where standard requests + BeautifulSoup are powerless, I use a bundle of pywinauto + OpenCV + OCR or Playwright + stealth + proxy.
I'm thinking about business, not code. Before starting, I always clarify why you need the data and who will use it. It often turns out that instead of complex parsing, the problem is solved more easily – for example, a direct SQL query or an API that the customer did not know about.
I work autonomously. For 3 years I have been conducting the project remotely, without daily calls and micromanagement. I myself understand someone else's interface, I myself find workarounds, I hand over the working result.
I honestly assess the timing and risks. If the task is technically impossible or is in a gray legal zone (for example, collecting personal data without consent), I will tell you about it immediately before I take an advance payment.
If you have a routine task of collecting, processing, or transferring data, write briefly what to do. I will tell you whether I will take, in what time and what result you will get.
What we need from you: Link to the target site or program name. Data collection list (which fields/files are needed). Example of the desired result (screenshot or blank Excel template). Access (login / password) if the site requires authorization. Protection information (if you know there is a CAPTCHA, Cloudflare or lock).
- Source code
- Documentation
- API integration
- Maintenance & support
- Система чек поинтов (Checkpoint/Restore)