Pakistan's PPRA publishes tenders on epads.gov.pk with no IT filter. Companies waste hours browsing hundreds of non-IT tenders manually.
A scraper + dashboard that scrapes PPRA daily, filters IT tenders using intelligent keyword matching, and displays them in a clean Bootstrap UI with search, filters, detail views, PDF downloads, and browser notifications for new tenders.
Reverse-engineering PPRA's undocumented HTML structure required many iterations — OpenCode helped me write robust BeautifulSoup selectors that survived site changes. The IT keyword classifier needed extensive tuning; OpenCode suggested the two-tier keyword system (title keywords + org keywords) that dramatically improved precision.
OpenCode was invaluable for the scraper's HTML parsing — I'd describe what I saw on the page, and OpenCode would generate the correct selector. It also taught me about APScheduler for production scraping, caching strategies to avoid redundant requests, and xhtml2pdf for PDF generation.
ML-based relevance scoring, email alerts, PWA support — leveraging OpenCode to implement features faster.
Great job creating a functional scraper and clean dashboard that address a real pain point for IT vendors. To strengthen the project, add a public code repository with documentation and consider a more robust classification model (e.g., lightweight ML) to improve precision. Also, include automated tests and deployment instructions to make the system easier to maintain and scale.