Latest Ensemble Learning for Multi-Label Image Classification Predicted the number of vehicles and signals in images using ensemble methods. Trained base models, including pre-trained MobileNet, a custom-built Vanilla CNN, and a Mean model baseline. Tableau Dashboard A set of Tableau dashboard developed out of interest and course work. Recurrent Neural Network (RNN) for Weather Prediction Built a custom RNN for predicting temperature based on historical weather patterns, achieving 91.19% accuracy (class top performance). Conducted feature engineering with time-series analysis techniques to handle sequential data. Software Development Creating Libraries for Boostrapping, resampling techniques, with Bioscience concept for research scientist, students and professor Thesis Deep Learning - Image Recoginition Image recognition-based Deep learning model for automated defect detection removing the manual inspection LLM Recommender Application Using APIs of Github/Hugging face/OpenAI/Twitter API generated structure DB with text embedding for recommending the appropriate LLM for the user prompt using vectore search Course Project Generative Adversarial Network (GAN) for Fake Image Detection Developed a multi-modal GAN combining image and text features for detecting fake data, achieving 97.59% accuracy (class top 1%). Utilized ResNet and BERT for robust image and text feature extraction. Autoencoder for Denoising Dataset Designed an encoder-decoder architecture to denoise data, achieving class-top performance with a 0.003 MSE. Leveraged MLFlow for hyperparameter tuning and achieved significant noise reduction and data reconstruction. Convolutional Neural Network (CNN) for Image Classification Leveraged MobileNet architecture for classifying COCO dataset images to detect the presence of people, achieving 92% accuracy (class top 1%). Implemented data augmentation and optimized for computational efficiency. Internships RAG - Data Reterival Search Engine A Sample of RAG whole system using Meta-Data Index/Semantic search engine (using Semantic Chunk) ML Model to Predict the Vulnerability of the php Application Developed a ML Model model to predict the vulnerability of an applications Assortment Modeling Developed Assortment Optimizer model in C++ using heurtics/Greedy Algorithm
