Machine Learning and Deep Learning enthusiast with hands-on experience in LSTMs, CNNs, and real-world forecasting tasks. Passionate about building AI-driven solutions, from language modeling to time-series prediction, with a focus on optimization and deploymen
Download CV
NLP
TensorFlow
OpenCV
YOLO
Matplotlib
Seaborn
NumPy
Pandas
Scikit-learn
XGBoost
LightGBM
Flask
Developed an LSTM-based language model on email data for generating coherent email-like text. Extracted emails via IMAP, preprocessed, tokenized, and trained a stacked LSTM model. Built a text generation pipeline predicting words from a seed phrase. Optimized with embedding layers, dropout, and softmax for improved performance.
Processed temperature data at 10-minute intervals (1,440 points over 10 days) for training and forecasting. Used a sliding window approach and baseline model (Test MAE: 2.62). Fully connected and CNN models failed. Improved forecasting with LSTM (Test MAE: 2.47), achieving a 5.72% improvement.
Welcome to FIT-U, an AI-powered fitness application designed to make your workouts smarter, easier, and more effective. Hosted on Render, this app leverages computer vision and pose estimation to track your exercises in real-time.
This project uses YOLOv8 for object detection combined with OpenCV for video processing and SORT to track vehicles in real-time. The goal is to detect and count vehicles in a video feed.
This project focused on predicting NYC taxi fares using XGBoost, successfully reducing the RMSE to $3.22. Enhanced computational efficiency while maintaining high prediction accuracy.
Predicting whether a mushroom is edible or poisonous based on physical characteristics. XGBoost achieved an exceptional MCC of 98.3%, showcasing its effectiveness in classification.