Projects
Few Shot Learning - Surrogate Learner
Improved an active learning surrogate model by adeptly updating the learner for every new data instance and yielding results, either rivaling or aligning with state-of-the-art techniques. Synthesized datasets to the above learner to enable it to perform different machine learning tasks such as NLP.
View Project
Mapease - Livestock Disease Tracking System
Streamlined monitoring of seasonal disease spread by tracking the movement of the livestock. Spearheaded a web application with a Spring Boot backend using JPA for data persistence and an Angular frontend featuring interactive maps built with Leaflet module in Angular.
View Project
File Share
Created a user-friendly file sharing Website using Firebase as the backend, lowering the file upload time to under 5 seconds. Implemented React JS to enhance the performance and responsiveness of the file sharing platform. Devised a QR code generator that automatically generates a unique QR code for each uploaded file, providing users with an easy and convenient way to download their files.
View Project
Covibot
Developed and implemented an advanced chatbot using the open source platform RASA, utilizing contextualized questioning and symptom analysis to accurately identify high-risk individuals with a success rate of 90%. Provided seamless integration with local health systems, enabling the chatbot to quickly provide contact information for nearby medical institutions, resulting in an average reduction of 30 minutes in obtaining medical assistance.
View Project
Early Detection Of Alzheimer's Disease
Trained a CNN model in Python, TensorFlow, and NumPy to diagnose different stages of Alzheimer’s disease from MRI reports. Constructed a dataset of 6400 MRI images after skull stripping the MRI images using the FSL library. Attained an accuracy of 80% when tested on a dataset of 2000 MRI images with a cross-entropy loss function
View Project
Slash - Online Shopping Application Scraper
Worked on a web scraping project implemented utilizing Beautiful Soup. Implemented secure user-based login with PostgreSQL and JWT. Successfully resolved challenges related to data consistency, resulting in a 20% improvement in data accuracy. Implemented a wishlist feature, making it easy for users to save and access search results. Added currency conversion, allowing users to view results in their preferred currency. Added a download option for search results to be downloaded in CSV format, enhancing utility.
View Project