AI-driven analytics system to nowcast and forecast PM2.5 air quality across Indian cities in real time. End-to-end ML pipeline integrating OpenAQ and Open-Meteo APIs, deployed as an interactive Streamlit dashboard with city-wise insights and anomaly alerts.
Customer segmentation engine analyzing purchase behavior with K-Means clustering and collaborative filtering for personalized recommendations.
Regression pipeline predicting urban taxi fares from trip and time-of-day features, benchmarked across multiple model families.
Cross-country analysis of obesity and malnutrition trends, surfacing the coexistence of both extremes within the same regions.
Paginated NASA API extraction pipeline into SQLite, surfaced through a Streamlit dashboard tracking near-Earth object approach data.
Systematic hyperparameter search (grid/random/Bayesian) across classifiers to optimize diabetes risk prediction.
Pending Archive
Multi-architecture CNN pipeline (ResNet50, MobileNet) for MRI-based tumor classification across 7,000+ scans, with augmented preprocessing.
Neural network built from scratch on the MNIST handwritten-digit dataset using TensorFlow — custom architecture, training, and evaluation.
Fuel-efficiency regression on the Auto MPG dataset using a deep learning model in place of classic sklearn regressors.