Aerosol Optical Depth Estimation

I built a pipeline to estimate Aerosol Optical Depth (AOD) by merging Sentinel - 2 satellite imagery with AERONET ground data. AOD is crucial for understanding air quality, climate, and public health. The work involved preprocessing cloud - masked satellite images and aligning them with AERONET AOD

Satellite Data Instance Segmentation for Agricultural Lands

I applied advanced instance segmentation techniques to high-resolution Sentinel-2 satellite images for agricultural lands. The project involved identifying field boundaries and analyzing crop distribution using image processing libraries and GIS tools. This work improves field detection accuracy for

Home credit : Credit Risk Model Stability

A model measured against feature stability over time. A comprehensive approach to constructing credit risk models.We employ algorithms like LightGBM and CatBoost, alongside ensemble techniques for robust predictions. Our pipeline emphasizes data integrity and feature relevance.

SepsisGuard

SepsisGuard, a machine learning system, predicts pediatric sepsis using machine learning with 0.983 AUC, addressing ICU data challenges. Built with Python/scikit-learn for clinical transparency and adaptability to real-time workflows.

RNN Music Generator

Hey! I've created a computer program that makes its own music! It learns from piano songs and can predict what notes come next. Plus, it can make longer pieces of music too. Isn't that cool?

BirdCLEF : Species Identification from Audio

The main goal of this project is to use machine learning to identify understudied Indian bird species by their calls. By analyzing continuous audio data, the project aims to create reliable classifiers that can work well even with limited training data.