Portfolio

How to Disappear Online

This Master thesis explores methodologies for digital erasure, focusing on data privacy, compliance with GDPR, CCPA, and AI regulations, and the use of machine learning to facilitate secure data deletion.

Open-Source Widn Integration with Blackbird.io

As part of my contributions to Widn, I worked on integrating it into Blackbird.io to enhance workflow automation using AI-driven text and file translation. This integration aims to streamline data workflows, optimize automated decision-making, and improve efficiency in real-world applications.

AI-Powered Virtual Fitting with Stable Diffusion

This project uses fine-tuned Stable Diffusion and LoRA models to create photorealistic virtual try-ons by applying specific clothing onto various body types with inpainting and data-driven AI training.

Truthful vs. Deceptive Hotel Reviews Classification

Utilized a variety of classifiers, including Linear Support Vector Classifier, Gradient Boosting Classifier, K-Nearest Neighbors Classifier, and Multinomial Naive Bayes Classifier to analyze hotel reviews. This project focuses on distinguishing between truthful and deceptive hotel reviews, showcasing my skills in handling complex classification tasks.

Helpful Reviews

Predict the helpfulness of Amazon Fine Food Reviews. The project integrates machine learning models like Naive Bayes and Logistic Regression with a Hugging Face transformer for advanced text analysis. It predicts review helpfulness based on textual content, combining traditional ML models with modern NLP techniques and utilizes sentiment analysis for a nuanced understanding.

Sentiment Analysis Project

Specialized in sentiment analysis, this Python-based project leverages NLTK and pandas for processing and evaluating text emotions. Key steps include data preparation, sentiment scoring, and detailed visualizations. The project effectively parses and analyzes sentiments of ideas and comments, providing insights into public opinion.