THE CURATED ARCHIVE
A repository of investigations into machine learning, adversarial security, and the intersection of traditional engineering with emerging AI paradigms.
Developed a multi-platform application teaching Spanish, English, French, and German using conversational AI. Evaluates user proficiency and dynamically adapts lesson plans using ChatGPT API and chain-of-thought reasoning.
Intelligent system combining a chatbot interface with regression modeling to extract property data and predict house prices in the Seattle market. Handled missing inputs using KNN.
Simulated distributed data parallel training using PyTorch to accelerate model training and ensure data privacy across multiple workers. Evaluated with an MLP on synthetic binary classification.
Evaluated four industry-standard datasets (KDD, UNSW-NB15, CICIDS, NSL-KDD). LSTM outperformed Naive Bayes in intrusion detection, achieving 99.97% accuracy on KDD Cup 99.
A robust, fully decentralized (P2P) chat application featuring end-to-end encrypted network architecture for confidential and optimal communication.
Advancing the detection of diabetic retinopathy by implementing and fine-tuning state-of-the-art Vision Transformer (ViT) deep learning approaches.
Applied deep learning techniques to improve speech recognition in noisy environments. Resulting CNN models struggled with increasing noise levels, driving GAN-based solutions.
"The project record is not a quiet place of storage, but a dynamic site of re-discovery where technology acts as the lens, not the filter."
EDITORIAL MISSION STATEMENT, 2024