• 

  model accuracy

• 

  check eligibility for loan and monthly emi corner

• 

  emi distribution over first 24 months and other features like loan eligibility etc.

• 

  emi advisor

Inspiration

Getting a loan can feel confusing—people don’t know if they’ll be approved, what the EMI will look like, or what to fix if they get rejected. We built CrediLume to make that process clearer and less stressful.

What it does

CrediLume helps you quickly check loan eligibility, estimate EMI and total repayment, and understand why the result looks the way it does. It also suggests practical steps to improve your chances. If you add a Gemini API key, it can give extra personalized guidance.

How we built it

We built a Flask web app with a simple UI (HTML + JS), connected it to a trained ML model (loan_model.pkl + features.pkl), and deployed it using Gunicorn on Render.

Challenges we ran into

The biggest issues were deployment-related (missing dependencies, model file paths, environment differences) and making sure user inputs don’t break the app.

Accomplishments that we're proud of

We’re proud that the app doesn’t just give a “yes/no”—it explains the decision, shows affordability (EMI/DTI), and still works even without the AI key.

What we learned

Real-world deployment is very different from running locally, and ML models need the right runtime libraries. We also learned that explainability and clear suggestions make the experience far more useful.

What's next for CrediLume

We want to improve the model, add more real-world inputs for better accuracy, and expand the advisor into more personalized, actionable financial planning.

Built With

• flask
• gemini-api
• gunicorn
• html
• javascript
• ninja-templates
• numpy
• python
• render
• scikit-learn
• scipy
• tailwind-css