/ WORK
MY WORKS
CRAFTING DIGITAL EXPERIENCES
with passion & code.
CreditSense AI
An explainable credit-risk model: XGBoost scoring with per-decision SHAP attributions, exposed as a FastAPI inference service backed by PostgreSQL. Built so every score can be explained, not just produced.
Flagship ML case study
- Python
- XGBoost
- SHAP
- FastAPI
- PostgreSQL
Delight Agro Traders
A production storefront for a real agri-input retailer in Khaga, Fatehpur district (Uttar Pradesh): a bilingual catalog, a WhatsApp-first inquiry flow, and a seasonal crop-advisory Knowledge Hub. Real, live software shipped for a real business.
Live in production
- Next.js
- TypeScript
- Supabase
- Vercel
UniCampus ERP
A full-stack College ERP built to replace the institution's existing third-party system with a custom, institution-owned platform. Scoped against a formal 15-section SRS across 11 modules, on a 32-week delivery timeline.
In active development — Week 4 of a 32-week build
- TypeScript
- Node.js
- React
- PostgreSQL
Titanic Survival Prediction
A hands-on ML playground built around the classic Titanic dataset: feature engineering and model benchmarking wired to live inference rather than a static notebook.
Interactive demo
- Python
- Pandas
- Scikit-learn
Credit Card Fraud Detection
A supervised classification project that flags fraudulent credit-card transactions. The core challenge is class imbalance — fraud is rare — so the pipeline uses SMOTE to resample before training and evaluation.
CodSoft internship project
- Python
- Pandas
- Scikit-learn
- SMOTE
Sales Prediction
A regression project that predicts product sales from advertising spend across channels — a compact, end-to-end look at fitting and interpreting a linear model.
CodSoft internship project
- Python
- Pandas
- Scikit-learn
Movie Rating Prediction
A regression project that predicts a movie's rating from features such as genre, director, and cast — an exercise in encoding categorical data and modelling a continuous target.
CodSoft internship project
- Python
- Pandas
- Scikit-learn
Iris Flower Classification
The classic introductory classification problem: predicting the species of an iris flower from its petal and sepal measurements. A compact, well-understood dataset used to practice a clean end-to-end classification workflow.
CodSoft internship project
- Python
- Pandas
- Scikit-learn
College Timetable Management System
A Java application for managing college timetables — creating and organising schedules — backed by a MySQL database and using Google OAuth for authentication.
Academic project
- Java
- MySQL
- Google OAuth
Old Portfolio
The earlier version of my personal portfolio: a single-page site introducing my work in machine learning and data science, with an animated hero, an interactive technology grid, and About / Skills / Projects / Education / Journey / Contact sections.
Live · previous personal site
- Next.js
- React
- TypeScript
- Tailwind CSS
- Framer Motion
- Vercel
