THE NEURAL
BACKBONE
See how SmartBank orchestrates data flow between NestJS, MongoDB, and our Predictive ML Engine to deliver real-time insights.
Data Flow Architecture
A high-performance pipeline built on NestJS and Prisma, connecting secure banking data with advanced machine learning models.
NestJS Core
The central nervous system. Handles Authentication (JWT), User Management, and orchestrates the data pipeline using modular architecture.
Predictive ML
Utilizing the Best Overall Model. Receives formatted customer data, runs classification models, and returns probability scores (YES/NO).
MongoDB Cluster
Stores raw customer profiles, historical prediction results, and campaign analytics in a scalable document structure.
Core Modules
Broken down into micro-services for scalability and maintenance.
Auth Module
JWT-based authentication with Access & Refresh token rotation. Secure Bcrypt hashing for passwords.
Users Management
Role-based access control (RBAC) handling ADMIN, STAFF, and USER permissions and CRUD operations.
Campaign Engine
JSON-based filtering logic ({'age': {'lt': 35}}) to target specific customer segments for batch predictions.
Customer Data
Handling bulk import (CSV/Excel) and export functionality. Validates attributes like job, marital status, and loan history.
Prediction Service
Manages the handshake with the external ML API. Stores prediction results (Probability YES/NO) linked to customers.
Analytics Aggregator
Compiles raw data into visual trends: Predictions by Job, Weekly Trends, and Conversion Overview.