Full Stack / AI / Systems
Smartloan BLMS
Loan management, with AI-assisted eligibility prediction.
- My role
- Full-stack development & AI loan-prediction integration
- Project period
- 2026
- Current stage
- Functional local prototype
The AI prediction interface combines applicant inputs with an eligibility estimate, risk indicator, EMI calculation and recommendations.
A problem worth
connecting.
Smartloan BLMS connects AI-assisted loan eligibility and risk prediction with the complete loan-management workflow. Loan officers review applicant finances, model outputs and repayment calculations in one place; administrators manage approvals, customer records, reports and configuration.
Built forLoan officers and financial-institution administrators
The work behind
the interface.
I designed and implemented the React interface, Express APIs, Prisma database models, authentication and role-based access. I also integrated the Python/Flask prediction service with the loan application workflow, bringing eligibility estimates, risk indicators, EMI figures and recommendations into the officer’s interface. The wider implementation includes repayments, document handling and administration.
A prediction needs its context.
AI-assisted loan prediction.
Applicant financial details pass from the React interface through the authenticated Express API to a dedicated Python/Flask service. The scikit-learn and joblib service returns an eligibility prediction and risk assessment alongside loan calculations and recommendations, so the officer can review the result in context.
Applicant context
- Monthly income
- Credit score
- Requested loan amount
- Loan duration
- Interest rate
- Existing monthly debt
Prediction service
Python · Flask · scikit-learn · joblib
Returned for review
- Eligibility prediction
- A model-assisted estimate of the applicant’s loan eligibility.
- Risk indicator
- A risk score or level returned by the prediction service.
- EMI & repayment context
- Monthly instalment and repayment figures displayed alongside the prediction.
- Recommendations
- Supporting response details presented for review with the application.
Eligibility and risk are model outputs. EMI and repayment figures are calculations displayed alongside them, giving the prediction context within the application workflow.
A closer look.
Actual screens from the project. Select a view or open it at full size.
The administrator dashboard connects portfolio summaries, loan trends and recent activity.
How the pieces connect.
Follow an action through the system.
- 01
Sign in
JWT authentication establishes the user’s role and permitted interface.
- 02
Apply
The React client sends authenticated customer and loan requests to Express.
- 03
Validate & store
Business rules are applied and Prisma reads or updates SQLite records.
- 04
Assess
Applicant data is sent to the Flask prediction service for eligibility outputs.
- 05
Manage
Approvals, repayments, notifications and reports reflect the updated loan state.
What the platform does.
Role-aware access
Administrators and loan officers sign in to dashboards and routes appropriate to their responsibilities.
Customer and application records
Customer profiles, KYC information, document uploads and loan applications stay connected through the approval process.
Loan configuration
Configurable loan types and interest rates support the application and EMI calculation workflow.
Repayments and overdue tracking
Payment schedules, EMI values, overdue balances and penalties provide a detailed view of an active loan.
AI loan eligibility & risk prediction
A dedicated Flask service connects scikit-learn model outputs to the loan application, returning eligibility, risk and recommendations alongside EMI calculations.
Operational visibility
Dashboard statistics, reports, notifications and activity logs help staff follow the state of the system.
This is where the learning happened.
The hard part.
Loan calculations, repayment status, access permissions and model outputs had to stay consistent across a JavaScript application and a Python prediction service.
How I approached it
I separated the React frontend, Express/Prisma backend and Flask prediction API. Shared request and response contracts and role-aware routes kept responsibilities clear while connecting the application, repayment and assessment workflows.
Tools behind the work.
- Frontend
- ReactViteReact RouterTailwind CSSRechartsLucide ReactAxios
- Backend & data
- Node.jsExpressPrismaSQLiteJWTbcryptMulter
- AI service
- PythonFlaskscikit-learnjoblib
- Communications
- NodemailerTwilio
Where it stands.
- A working local loan-management implementation across React, Express and Flask.
- Connected customer, application, approval and repayment workflows.
- An integrated eligibility service presented alongside loan calculations and recommendations.
- Responsive administrator and loan-officer dashboards.