Projects
Safeguarding AI
Author : Pending Publication
Project explores machine learning attacks' societal impact, advocates robust defenses, and aims to raise awareness about identifying and preventing these threats. It delves into various attack types, offers defense strategies, and envisions an AI future prioritizing safety, security, ethics, and privacy.
Kinase Inhibitor Classification
Trained a classifier on KinFragLib data to categorize fragments into seven labels, focusing on AP fragments. Identified unique fragments using dice similarity score, classified them using random forest and logistic regression models based on chemical fingerprints. Evaluated expert-curated descriptors' impact on models' performance in distinguishing AP fragment labels, capturing key features like lipophilicity, exit vectors, and aromatic rings.
Blender Created Scientific Image in "The Gut-Wrenching Effects of Cryptosporidiosis and Giardiasis in Children"
Collaborated with the University of Washington to create a scientific illustration in Blender 3.5.1, depicting an infected host showcasing Cryptosporidium and Giardia. Explored how these infections potentially disrupt gut microbiota balance, visualizing dysbiosis within epithelial cell layers via altered colored particles. Highlighted the impact of Giardia duodenalis cysteine proteases in breaking down mucins, crucial components of intestinal mucosal layers. This visual representation aligned with histological evidence, as detailed in the MDPI article "The Gut-Wrenching Effects of Cryptosporidiosis and Giardiasis in Children."
An Exploration of Virtual Screening Techniques for Small Molecule Inhibitors for COVID-19
In this project different packages such as DeepChem, DeepPurpose and other published
deep learning architectures were explored in order to rank binding affinities of small molecules to COVID-19. The rankings of the models have been validated in the nanotechnology lab at UCF. Models are being further optimized. The goal of the project is to find a material that can capture and kill COVID-19. Models include GCNN and DNN. Further research is exploring how to leverage PauliNet information and also research in Hypergraph representation for deep learning
Simulating Fear within Virtual Reality
Crafted a fear simulator in virtual reality by leveraging non-linear sounds, eerie motifs, and realistic visuals. Demonstrated the simulation's success through a Google Cardboard-based VR maze game, effectively evoking fear by engaging user senses.
Air Pollution Detection System using Edge Computing
Explored IoT for real-time pollution detection via Arduino and Raspberry Pi in Tacoma. Led web development and authored a conference paper. Prototype aimed for use in Azores Islands research. Next phase involves app development for field data recording. Skills gained: IoT fundamentals, conference paper writing, Azure basics, Ionic app, JavaScript, Angular.
Project RubiX
As Vice-President of UWT's Mathematics Club, our aim is fostering an engaged math community while demystifying math fears. Through Project RubiX, we merge engineers, computer scientists, mathematicians, and artists to craft and showcase a robot solving a 3x3 Rubik's cube. This interdisciplinary endeavor involves registered student organizations, fostering community while navigating communication challenges in diverse projects. Our presentation navigates Project RubiX's timeline, challenges faced, and contrasts initial goals with achieved milestones. Read more here
