This project explores combining the flexibility of Large Language Models (LLMs) with the statistical rigor of Conformal Prediction, which is used to create a statistical model of the per-sample uncertainty in the LLM's prediction.

Chat with a PDF - A Working Demo

Investigation of some data and many papers cultiminating in a 'Grand Rounds' presentation on the topic of seasonal variation in TSH levels and implications for reference ranges.

QI Project where we had to request data from cerner (with help), perform data analysis as to usage and utility, and then present recommendations at semester's end.

Data Analysis of mycobacterial cultures in osteoarticular specimens

Commentary on contextualizing model performance with that of naive predictors in sepsis prediction.

Leverages the logical circuit decomposition introduced in my previous project to interpret the AI classifiers for ECG signals.

Introduces a novel representation of AI classifiers as logical circuits through a min-max ReLU layer representation to improve interpretability and explainability.

A 'support vector' like approach to complexity bounding of neural networks in deep learning

Exploring the ability of GANs to learn causal structure

Independently designed and built high-end computer workstation for AI research applications (10k budget)

Histopathologic Pixel-Level Segmentation Competition

python GUI for 3D neuron segmentation

Best-in-Class lipreading model Classifies Spoken Words without Audio

Intro to Coding with Biomedical Project