Purpose & Vision

TheBudget.Watch is designed to tackle corruption and irregularities in public sector procurement — beginning with road construction tenders in Morocco. By analyzing open tender data from platforms like TGR, the system detects suspicious patterns using AI models and flags potentially fraudulent bids. Long-term, the project aims to support journalists, NGOs, and government agencies in proactively monitoring public spending abuse.

Features

  • Kafka-based pipeline for ingesting and processing tender data
  • AI model trained using Scikit-Learn to detect anomalies in public road contracts
  • Interactive dashboard for risk-level visualization
  • Document parsing and tagging (future roadmap)
  • Multi-stakeholder access model (citizens, NGOs, journalists)

Tech Stack

React
Tailwind
Node.js
Python
Scikit-Learn
Kafka

My Role

  • Defined the architecture and data ingestion pipeline (Kafka + Node.js)
  • Trained and deployed fraud detection models using scikit-learn
  • Developed the frontend for visualizing tender risks
  • Built data parsers for CSV, Excel, and JSON tender formats
  • Maintained end-to-end integration and MVP deployment

Impact

  • Raised awareness of public procurement irregularities through data
  • Created a functional prototype with real public data sources
  • Contributed to open civic tech by enabling data transparency
  • Attracted interest from journalists and civic technologists