Deep-Learning Stack

A repository encompassing projects on Deep Learning, Transformers, Open AI, Image Classification, Object Detection, Reinforcement Learning, Computer Vision, NLP, and Generative AI, including RAG and LLM. This repository features comprehensive projects from all these topics.

Home-Credit-Credit-Risk-Model-Stability Kaggle-Competition

A Kaggle competition hosted by Home Credit helps assess potential clients' default risks, enabling consumer finance providers to accept more loan applications. This may improve the lives of people who have historically been denied due to lack of credit history.

Smarter Medicare Planning Through Cost Prediction

This project focuses on building an accurate model to predict Medicare costs, aiming to alleviate healthcare financial burdens and empower insurers and patients to make more informed decisions. Comprehensive data exploration was a critical component, revealing significant patterns and trends.

Large-Scale PDF Analysis Leveraging RAG and LLMs for Enhanced Machine Learning Literature Insights

The project enhances ML research accessibility using RAG approaches and LLMs like Llama, Mistral, and Gemma. It facilitates discussions and streamlines literature reviews by querying PDF papers. Deliverables include an ML-Chat app for providing insights and engaging with ML literature.

Real-Time-Emotion-Detection-with-Kafka-and-ML

Developed a real-time emotion detection system using Apache Kafka, Spark Streaming, and ML models like Naive Bayes, SVM, and Random Forest. It processes Twitter data to identify emotions such as anger, joy, and sadness, providing immediate feedback for sentiment analysis, and personalized marketing.

Predicting Air Quality for Health Risk Assessment: A Machine Learning Approach

Created a hybrid model combining Time Series Analysis and Machine Learning to forecast AQI. ARIMA is used for forecasting, with SVM, Decision Trees, and XGBoost enhancing accuracy. Cross-validation and ablation studies ensure robustness, demonstrating consistent performance.

Taxi-Trajectories-A-Predictive-Approach-to-Forecasting-Fare-Duration-and-Demand-Dynamics-in-NYC

The project "Taxi Trajectories" aims to optimize NYC taxi operations for the year 2023 by predicting trip duration, fare, demand and forecast using machine learning models. It involves extensive data preprocessing, feature engineering, and model evaluation.

CardioRisk-Unraveling Patterns for Early Diagnosis Using Machine Learning

Developed predictive models for early CVD detection and risk assessment using the UCI heart disease dataset. Utilized PCA for data preprocessing and models like XGBoost, DT, SVM, RF with Linear Regression to enhance global cardiovascular health by enabling personalized healthcare and early detection

Winehaven-E-Commerce-Application

It offers an intuitive GUI for users to browse, reserve, and purchase wines while managing their collections. Features include secure user registration, detailed product catalogues, tasting reservations, advanced search and filtering, a shopping cart, and streamlined checkout.

Performance Analysis of Machine Learning Models using Academic Data Sets

Published a paper- "A survey on the efficiency of Machine Learning models on the academic performance of students" ISBN: 978-93-80506-20-3 Prathibodh, a National Conference on “Shaping the future: Digitize India! Start-up India