MY PROJECTS
Explore my Artificial Intelligence projects, where I unravel insights and solve real-world challenges through the power of data.
NVIDIA Hackathon: GPU Accelerated Predictive Modeling
-
Developed a high-performance predictive model for the NVIDIA Spooktacular Data Science Competition at ODSC West 2024, working with an extensive 9.52 GB dataset containing ∼ 11 million samples and 106 features, achieving 8th place among 45+ teams.
-
Leveraged NVIDIA's RAPIDS libraries (cuDF, cuML) to optimize data processing and model training on an A100 GPU, reducing run time to 1 minute and 55 seconds through advanced GPU computing techniques.
Shooting Mastermind - RAG Based Chatbot
I developed a RAG-based chatbot to simplify navigating complex shooting sports regulations. By gathering handbooks from sources like the ISSF, the chatbot provides accurate and quick answers on rules such as weapon specifications and equipment standards. The tech stack includes GPT-4.0 Mini for prompt responses, GTE-base for text embeddings, Pinecone for vector search, and is deployed on Heroku with a CI/CD pipeline managed via GitHub Actions.
Molecule Recommendation
Contribution Towards Chemical Discoveries: Introducing Our Innovative Molecule Recommendation System!
Our project is dedicated to designing a hybrid chemical recommender system. This system leverages a combination of algorithms, including collaborative filtering, content-based approaches, Graph Neural Networks (GNNs), and autoencoders. Our goal? To make it easier for scientists to find new chemicals they didn't even know existed, allowing them to make discoveries.
​
AI Content Approval
Tamu Datathon Marky Challenge
(2nd Place)
MarkyChallenge is a data science project developed for Tamu Datathon, focusing on predicting user approval of AI-generated content. By analyzing both textual and visual data, our project leverages machine learning techniques to provide valuable insights into user preferences, offering potential enhancements to content creation and recommendations
TAMU Datathon 2o23
Exploratory Data Analysis on CHD Dataset
In this exploratory data analysis (EDA), I delved into a dataset sourced from Kaggle, which is focused on understanding the risk factors associated with coronary heart disease (CHD). Recognizing and forecasting CHD risk is crucial in public health, as it enables early intervention and minimizes morbidity and mortality. By analyzing this dataset, I aimed to shed light on the complex relationships between factors such as smoking habits, health variables, and CHD risk, particularly emphasizing the impact of smoking on cardiovascular health.
DATABASE INTERACTION USING AUTOMATIC SPEECH RECOGNITION
Constructed a multilingual database system enabling users to interact in their native language. Used Automatic Speech Recognition for spoken-to-text conversion and Natural Language Processing (NLP) to perform lexical, syntactic, semantic, and pragmatic analysi
Fashion MNIST
In this collaborative project, I worked alongside two team members to develop a clothing classification system using the FashionMNIST dataset. We aimed to create a robust classifier that can identify different types of clothing items with high accuracy. We designed and executed training and validation strategies, aiming to achieve the highest accuracy in classifying clothing items. Our efforts encompassed various aspects of machine learning, data preprocessing, model architecture, and hyperparameter tuning.
TD Hospital Exploration
TD-Hospital-Exploration is a data-driven project designed to uncover vital patient survival factors and bolster decision transparency at TD Hospital. Through meticulous analysis of extensive structured patient data and advanced machine learning techniques, we strive to offer actionable insights.
