Projects

New Jersey clinics have to send patient charity care data to the state in one very specific format. Most do it manually.

During my time with Jasfel Analytics, I built the platform that does it for them. Upload your data, and it comes out clean, validated, and ready to submit in seconds.

From there it grew past compliance. I trained a predictive model that flags which upcoming appointments are likely to be missed, so clinics can fill those slots instead of eating the loss.

Then I gave the analytics dashboard a voice: a lightweight LLM running locally on the clinic's own data. Staff ask a question in plain English and get the numbers and context back on the spot, without waiting on an analyst and without a single patient record leaving the building.

Measuring how well someone moves usually requires an expensive motion capture lab. My research asked: what if a regular camera plus small wearable sensors could do the job?

I built the pipeline that makes it work. It watches video of a patient moving, tracks their body automatically, syncs that with the sensor readings, and turns it all into movement quality scores a clinician can use.

Did liver disease in kids get more common after the pandemic? That's the question this study set out to answer, using national health survey data covering thousands of children.

As a co-author, I handled the data side: cleaning huge messy datasets and making sure every number in the analysis could be trusted. The study was accepted for an oral presentation at DDW 2026, one of the biggest GI conferences in the world.

This study looked at whether where someone was born and their background relate to how severe liver scarring gets. Questions like that matter for catching disease earlier in the right communities.

I was second author. My job was the analysis: cleaning the data, checking its quality, and building a statistical model that stays fair even when the survey data over-represents some groups. Presented at AASLD 2025, the top liver conference in the country.

For a year, I taught middle schoolers how AI actually works, not just how to use it. We built and trained real models together, including a chatbot the students made themselves.

If you can explain how a language model works to a room of 12-year-olds, you can explain it to anyone. Easily one of the most fun things I've done.