Computational genomics @ the Gao Lab

Metis, the titan of wisdom and Zeus's trusted planner in Greek mythology inspires our platform's name. We aim to be the modern-day Metis for every college student, providing lightning-fast personalized, and insightful advising plans with AI.

Used Pseudo Hilbert Space-Filling Curves to map the bytes of Portable Executable files into 2-dimensional matrices. Matrices were used to train a CNN to classify if the file was malicious or not. The final CNN model achieved an accuracy of 90.32%, with a 0.92 AUC score, 0.9536 Specificity/Precision, 0.9047 F1 Score, and a 0.8605 Sensitivity/Recall. The model was deployed to a web application demo where users can upload a file and get a probability of whether their file is malicious or not.

Modeled protein structure by using DeepMind’s AlphaFold machine learning model to identify mutated STAT3 proteins and their effect on the development of Large Granular Lymphocyte Leukemia

Created a GAN (generative adversarial network) in Tensorflow to generate synthetic packet data for network simulationCreated an end to end traffic generation system with NLP (natural language processing) to take keywords from a user profile and browse the internet based on the profile's perceived interests

Created a mail server with an Amazon EC2 instance.

Working on a CNN architecture to classify Alzheimer's progression

Refereed multiple soccer games of kids ranging from 7 to 16 years old.