Ted Talk Views Prediction

A Regression model was developed to predict number of views of a TEDx event based on a dataset about past TED events. Various supervised learning algorithms like Linear Regression, Decision Tree were applied along with regularization and hyperparameter tuning

EDA-Hotel Booking

The project is based on performing EDA on a dataset containing information about two type hotels, City and Resort and their booking details. EDA was performed on the given dataset to understand and infer the booking trends and the various factors affecting it.

Hindi-Vowel-Consonant-Classification

Task: In a given image of an Hindi alphabet classify the vowel and consonant

Handwritten digit generation using DCGAN

A Pytorch implemetation of DCGAN on the MNIST dataset to generate fake handwritten digits

Google Satelite Image to Streetmap Image translation using Pix2Pix GAN

Pytorch implementation of Pix2Pix model from scratch to generate street map images of a corresponding satellite images

Video Subtitle and Summarizer using Deep Learning

Developed a web application for video summarization using Flask framework which would transcript and summarize the videos. Used OpenAI's Whisper model for transcribing and Fine-tuned BART model for summarization

Medical Chatbot

It is a user friendly tool to ask for any disease related queries which can be helpful for common people. Developed application using Langchain framework and Meta Llama2 quantized model

MathHelp

It is a user friendly tool for solving any mathematical problems. User can ask any maths related question and the tool will return the answer Developed the application using Langchain framework and GPT 3.5

Text Classification with BERT

Finetuned BERT model on a custom dataset containing tweets to classify their sentiment. Each tweet can be classified into 8 sentiments and can have more than one categories For e.g a tweet can be both 'joyous' and 'optimistic'

Chest-Cancer-Classification

An End to End Chest Cancer classification app created to take input image of chest and classify if cancer is there. VGG16 model was used to classify the images. MLOPs tools DVC, MLflow were used to track ML pipelines in the project. Deployed project using CI/CD principles on AWS EC2 instance