Delivered·Dashboards & intelligence ·SaaS & software·usa-east

Confidential engagement

Machine learning-based procurement fraud detection engine for large enterprises and government agencies, identifying suspicious purchase orders, vendor fraud patterns, bid manipulation, and kickback schemes using behavioral anomaly detection.

Next.jsTypeScriptPython/scikit-learn for ML modelsReal-time data ingestion from ERP systems (SAP, Oracle)Supabase for transaction storageGraph database for vendor relationship mapping

Confidential engagement · usa-east

What we built

01Real-time PO anomaly detection (unusual amounts, vendors, timing)

02Vendor network analysis (shell companies, related-party connections)

03Bid collusion detection across tenders

04Price variance analysis and outlier identification

05Buyer-vendor relationship mapping (conflict of interest detection)

06Historical fraud pattern matching

07Risk scoring for each PO with explainability

08Alert routing to compliance/internal audit teams

Technical stack

Next.jsTypeScriptPython/scikit-learn for ML modelsReal-time data ingestion from ERP systems (SAP, Oracle)Supabase for transaction storageGraph database for vendor relationship mappingExplainable AI (SHAP) for fraud reason transparencyTime-series anomaly detection algorithms

More work

SaaS & software

NileRoute OS

Live · Morocco / International

AI systems

Evo2 Variant Intelligence

Delivered · International

SaaS & software

SignalsFrame

In development · International

Want to build something like this?

Have a system, product, campaign, or visual experience that needs building?

Start a project