SafeLens - Adversarial Attacks on ONNX AI Models

In this project, the security vulnerabilities of the deployment of AI models on the web browser are analysed. For this, a convolutional neural network is trained on the TinyImagenet-200 dataset for image recognition and deployed on the browser. Adversarial attacks target the accuracy of inferencing.

Manifold-Based CCA for Gene-Drug Sensitivity

This is my latest project in math - here, the numerical stability of geometric optimization is used to analyse the relationship between the expression of certain genes in breast cancer patients and their correlated sensitivity (or not sensitivity) to most used drugs.

LLMCoder: A Feedback-Based Coding Assistant

LLMcoder is a coding assistant addressing a common problem found in many completion tools -such as Github Copilot: most of them do not use the latest knowledge about packages, APIs and environment information. LLMCoder has proven 25% more accuracy over Copilot. Feel free to visit the repository.

Homomorphic Polynomial Public Key Cryptography

This work focuses on understanding the mathematical foundations behind these methods and their practical applications, especially in a post-quantum world. In the future, I aim to contribute to developing secure cryptographic solutions that are future-proof and robust against emerging threats.

CCA on Multivariate Reduced-Rank Data Analysis

This is a series of projects developed during the summer 2024 on continuation to the PCA project of last year. It aims to construct the mathematical framework of CCA for data visualization, and later apply it to optimization on matrix manifolds, which will be used through the Cholesky retraction.

Smart Play: A Checkers AI with Minimax Optimization

During the winter semester 22-23, I followed a course algorithm engineering where we were empowered to propose solutions for well-known game theory problems and explore the aplications of MCTS. In my case, I opted for the implementation of the Minimax algorithm for checkers.

Utilizing PCA for Osteoarthritis Diagnosis

This project explores Principal Component Analysis (PCA) from both theoretical and practical perspectives. I focus on its application to analyzing the composition of Synovium tissue to detect the presence of Osteoarthritis in patients.