Pittsburgh, Pennsylvania
Featured Project

RedViz: LLM Red-Teaming Visualization Framework

December 15, 20257 min read
Streamlit dashboard for probing LLM safety across 8 languages and 9 harm categories with live jailbreak testing and attention analysis.
Inside the case study
What RedViz IsWhy We Built ItThe DataFour Dashboard Modules

EqRAG: Financial Stock Prediction with Fine-Tuned LLMs

LoRA fine-tuning on 7B math LLMs for Dow 30 stock prediction. Jumped from 12% zero-shot to 51.7% accuracy by training 0.19% of parameters.
GAN-Based Synthetic Portrait Generation with Pix2Pix

GAN-Based Synthetic Portrait Generation with Pix2Pix

Three GAN architectures (DCGAN + Pix2Pix) trained to generate anime faces from video frames. 28.5 dB PSNR, 0.91 SSIM.
Telecom Customer Churn Prediction: ML Pipeline with Neural Networks

Telecom Customer Churn Prediction: ML Pipeline with Neural Networks

ML pipeline comparing five classifiers on 7,043 telecom customers. Neural network hit 0.83 AUC. Contract type was the strongest churn signal.

News Sentiment Analysis with BERT and Web Scraping

NLP pipeline scraping Google News and classifying sentiment with fine-tuned BERT. Includes aspect-level entity extraction.
Skin Burn Detection and Classification with YOLOv5

Skin Burn Detection and Classification with YOLOv5

YOLOv5s classifying burn severity into 3 levels on 14,044 medical images. Transfer learning from COCO. Medically validated annotations. Top 5 finalist at the TiE Global Hackathon.

Credit Card Fraud Detection

Fraud detection on 284,807 transactions with 0.172% fraud rate. Random Forest achieved 99.96% accuracy. Built without resampling.
FIFA Ultimate Team Squad Builder: ML Optimization with Random Forest & MILP

FIFA Ultimate Team Squad Builder: ML Optimization with Random Forest & MILP

Scraped 15,000+ FIFA player cards, predicted goals with Random Forest, and optimized squads with MILP. Acceleration beat shooting stats.