RESEARCH

Computational Reasoning Group

Jun 2024 - Present

Scraped data using Selenium of 200,000 UF engineering alumni from LinkedIn for career analysis
Used and compared different LLM’s performance in automatically labelling the LinkedIn data in 7 categories
Tuned prompts to achieve over 90% accuracy in all categories and performed trend analysis on resulting labels

Jan 2024 - Present

Employed TensorFlow to create a Wasserstein Generative Adversarial Network (WGAN) to supplement
COVID CT scan data by creating artificial images with extremely high stability and reproducibility
Trained WGAN model on over 100,000 training images using a supercomputer and SLURM batch jobs
Evaluated WGAN performance with FID, LPIPS, and MMD, demonstrating the model’s high fidelity
Compared performance to other popular GANs to demonstrate the effectiveness and potential of our model

Sep 2023 - Present

Finetuned several temporal video segmentation PyTorch models and compared their accuracy and analysis time
Utilized UF’s supercomputer along with SLURM scripts to allocate hardware to train video segmentation models
Made an iOS app with Swift and Apple Vision to do edge analysis on videos and send them to Firestore
Made a duplicate Android app using Kotlin and Mediapipe