Projects

Machine learning projects spanning production systems, recommendation engines, and predictive modeling.

Skinwise: AI-Powered Skincare Recommendation System

Sentiment analysis pipeline on 581K product reviews achieving 86% accuracy with 46% improvement in negative feedback detection.

Tiny Transformer for Next-Token Prediction

Built transformer from scratch for Shakespeare text generation, achieving 51% perplexity improvement through systematic hyperparameter optimization.

Bank Marketing Campaign Optimization

Ensemble ML models achieving 36.9% F1-score on severely imbalanced dataset (11.3% positive class) with 22% improvement over baseline.

Heart Disease Prediction: Comparative Analysis

Compared Naive Bayes and regularized regression models, achieving 83.5% accuracy with Ridge regression on UCI medical dataset.

Exploratory Data Analysis & Visualization

Advanced data visualization techniques using R to uncover insights across multiple domains—from housing market analysis to quality-of-life indicators and geo...