Double-Funding Prevention
Industry
Government
Client
Finance Ministry of Lithuania
Team
PILOT
Year
2025
We were hired to create an automated system that prevents double financing of public projects by monitoring all funding sources daily and alerting to potential overlaps before they become problems.
Challenge
Multiple IT systems managing various (national and EU) funding sources in Lithuanian public sector projects created significant risk of double financing, where the same expenses could be funded more than once. Manual oversight was insufficient to catch overlapping funding across different state systems, leading to potential misuse of public funds and accountability issues.
Approach
Built a national platform that performs automated daily cross-system checks (entity codes, activities/objectives, invoice/account numbers) to flag potential double-financing.
Provided a case-management web UI so administrators can review overlaps, add notes, resolve cases, and generate reports; added email alerts with summarized overlaps.
Implemented secure access with Microsoft Entra ID SSO, role-based access control, and AES-256 encrypted data transfers (GDPR-compliant).
Delivered on a scalable microservice architecture: Python (Flask/Django) + SQL Server back-end, Next.js/React front-end; designed for future ML/LLM upgrades.
Architecture/Backend
Python (Flask) with SQL Server; Next.js; scheduled sync across external state IT systems.
AI/ML models
Created grounds for integration of LLMs to improve double-financing detection.
Infrastructure/Deployment
Microsoft Entra ID authentication with RBAC; AES-256-encrypted data transfers.
Key results
Replaced labor-intensive manual reconciliation with automated daily detection and unified reporting, significantly reducing administrative burden.
Reliability & scale at launch: supports up to 50 concurrent users with ≥96% uptime targets.
Governance & security by design: SSO + RBAC and encrypted transfers to protect sensitive financial data while centralizing oversight.
Enhanced financial accountability through systematic tracking and reporting capabilities, allowing administrators to efficiently review and address funding overlap issues.
Scalable architecture positioned for future machine learning integration to improve detection accuracy and reduce false positives
