How pushkar1805/ai-fintech-risk works
RiskLedger AI is a synthetic-data underwriting and fraud-risk prototype. A user enters customer and cash-flow signals in the application workspace; the FastAPI assessment endpoint validates the application, computes credit, behavioral and fraud risks, then returns a decision, recommended limit and reason factors while persisting assessments. Separate dashboard, customer and model-performance views expose product data. The README also documents synthetic-data generation and model training; their runtime wiring is not established by the sampled excerpts. The README describes SQLite, while backend configuration requires a DATABASE_URL, so the concrete configured database is uncertain.
Main entry points: main.py FastAPI endpoints · api.ts API client · db.py Database access · train_models.py Model training
Components 20
Web application
- Application workspace
frontend/app/application/page.tsx - Dashboard
frontend/app/page.tsx - Customer view
frontend/app/customers/page.tsx - Fraud monitor
frontend/app/fraud/page.tsx - Model metrics
frontend/app/models/page.tsx - API client
frontend/lib/api.ts
API service
- FastAPI endpoints
backend/app/main.py - Application validation
backend/app/schemas.py
Risk intelligence
- Risk assessment
backend/app/main.py - Alternative credit model
backend/trained_models - Fraud and anomaly models
backend/trained_models - Behavior and decision rules
backend/app/main.py - Model training
backend/ml/train_models.py
Data and models
- Synthetic data generation
backend/ml/generate_data.py - Customer and assessment records
backend/app/models.py - Database access
backend/app/db.py - Configured database
- Synthetic CSV data
backend/data - Model metrics artifact
backend/trained_models/metrics.json
Other components
- Underwriting user
Connections 20
- Underwriting user → Application workspace: enters signals
frontend/app/application/page.tsx - Application workspace → API client: requests assessment
frontend/app/application/page.tsx - API client → FastAPI endpoints: posts assessment
- FastAPI endpoints → Application validation: validates input
backend/app/main.py - FastAPI endpoints → Risk assessment: runs assessment
backend/app/main.py - Risk assessment → Alternative credit model: predicts credit risk
backend/app/main.py - Risk assessment → Fraud and anomaly models: scores fraud risk
backend/app/main.py - Risk assessment → Behavior and decision rules: computes decision
backend/app/main.py - FastAPI endpoints → Database access: gets database session
backend/app/main.py - FastAPI endpoints → Customer and assessment records: creates and queries records
backend/app/main.py - Database access → Configured database: connects to
backend/app/db.py - Customer and assessment records → Database access: uses base metadata
backend/app/models.py - Dashboard → API client: requests dashboard
frontend/app/page.tsx - Customer view → API client: requests customers
frontend/app/customers/page.tsx - Model metrics → API client: requests metrics
frontend/app/models/page.tsx - FastAPI endpoints → Model metrics artifact: reads metrics
backend/app/main.py - FastAPI endpoints → Synthetic CSV data: reads customer data
backend/app/main.py - Model training → Alternative credit model: persists selected model
- Model training → Fraud and anomaly models: persists risk models
- Synthetic data generation → Synthetic CSV data: writes datasets
backend/ml/generate_data.py