Tender Finder: Automated Procurement Crawling and Classification Engine

Every day, hundreds of tender notices are published across thousands of fragmented government e-GP portals, corporate procurement websites, daily print newspapers, and NGO portals. Businesses face significant challenges consistently monitoring these sources, often missing critical bidding opportunities due to delayed discovery, poor categorization, or insufficient response times. For Tender Finder, the engineering hurdle was building highly resilient, distributed web crawlers to scrape unstructured HTML and PDF attachments, creating an accurate automated machine-learning classifier to standardize categorization, and designing a high-throughput notification broker capable of distributing thousands of personalized alert emails and SMS messages daily without queue congestion.

Cloud House Technologies developed a robust, automated GovTech portal. We engineered a scalable web-crawling cluster using Python Scrapy and Selenium, orchestrated via Celery tasks to monitor e-GP databases, digital newspaper archives, and corporate portals. We designed an intelligent classification engine that pre-processes incoming tender listings, extracting keywords to automatically assign categories (e.g. ICT, Civil Works, Medical Supplies). The core Laravel application stores these in a high-performance PostgreSQL database. Subscribers configure personalized alert profiles on a reactive frontend built with TailwindCSS. Finally, we built a Redis-backed queue system that matches daily tenders against user profiles, dispatching instant, personalized email, SMS, and push notification digests.

• Over 40,000 active procurement subscribers receiving customized daily alerts.
• 97.8% classification accuracy achieved through refined NLP processing, eliminating manual sorting.
• Bidding preparation lead time increased by an average of 5 days for subscribing enterprises, resulting in a 22% average increase in proposal success rates.
• Manual search overhead reduced to zero, saving organizations dozens of administrative resource hours weekly.