Projects
Builds
Applied machine learning projects built and evaluated end to end. Each entry describes the problem it solves and how it was tested. Filter by domain below.
-
ScrubCheck: An NCCI claim scrubber that runs before submission
It checks a claim's procedure codes for NCCI bundling and unit cap edits before submission and explains each flag in plain language.
-
SourceFetch: Grounded Q&A over billing & coding policy
A retrieval augmented assistant that answers billing and coding questions only from the source manuals. It cites each passage and says so when the answer is not there.
-
Evaluation and Error Analysis on a Multitask Review Triage Capstone
The evaluation and error analysis workstream for a graduate capstone at GWU: a BERT triage classifier with five heads measured down to its real confusion matrix and the boundaries where it fails.
-
Glossary Flashcards: An image to Anki deck generator
An older web app that turns book glossary photos into flashcards for Anki. Remastered with Claude's vision API.
-
Free-Tier Benchmark: Assessment framework for free LLMs
A benchmark assessing the quality of output from three free-tier LLMs (Perplexity, ChatGPT, and Gemini) across three notebooks and 26 hand annotated cells.
Not all work is public. For questions about a project, get in touch.