Portfolio

This project focused on classifying text into three categories: hate speech, offensive language, and non-offensive using deep learning models. The team implemented and compared various models, including LSTM and Bidirectional LSTM, to determine the most accurate approach for detecting hate speech.

My key contribution was developing and optimizing the Bidirectional LSTM model, handling data preprocessing, model training, and evaluation. After testing multiple models, we selected the best-performing one based on accuracy and validation results.

This project highlights the importance of AI in fostering respectful online communication and combating hate speech, contributing to safer digital spaces.

Amazon Sentiment Analysis
This project focuses on analyzing customer sentiment in Amazon product reviews. The goal is to classify reviews as positive, neutral, or negative using deep learning techniques.

To solve this, Natural Language Processing (NLP) is applied to preprocess the text, including tokenization, cleaning, and embedding words with GloVe for better context understanding. A Bidirectional LSTM model is used to capture long-term dependencies in text, improving sentiment classification accuracy.

This solution can help businesses gain insights into customer feedback, enhance product recommendations, and improve user experience by identifying trends in customer sentiment.

Movie Recommender System 🎥
This project builds a hybrid movie recommender system that personalizes movie suggestions using content-based filtering and collaborative filtering techniques.

The system analyzes movie descriptions using TF-IDF Vectorization and measures similarities with Cosine Similarity for content-based recommendations. It also applies User-Based and Item-Based Collaborative Filtering using KNN to enhance recommendations based on user preferences.

By combining these techniques, the system improves recommendation accuracy while ensuring diversity. It processes movie data efficiently with pandas and NumPy and visualizes insights using Matplotlib and Seaborn.

This project focused on object detection to identify trash in images using deep learning models. The team implemented and compared CNN and YOLO models to determine the most accurate approach for detecting waste.

My key contribution was developing the CNN model, handling data preprocessing, model training, and evaluation. After training both models, we selected the best-performing one based on accuracy and efficiency.

This project highlights the power of AI in environmental sustainability, helping automate waste detection for cleaner surroundings.

This project focused on detecting whether individuals are wearing helmets in images using deep learning models. I implemented the YOLO (You Only Look Once) model to accurately detect and classify helmet usage in real-time.

My key contribution was developing the YOLO model, handling data preprocessing, model training, and evaluating its performance. After training and testing, YOLO was selected as the best-performing model based on accuracy and inference speed.

This project demonstrates how AI can be applied to safety applications by automating helmet detection to promote safer work environments.