Writing documentation is a major burden on healthcare professionals, especially physical therapists. We are creating a tool called ScribePT, which automates the process of creating physical therapy documentation (SOAP notes), based on a recording of the session between the PT and patient. It is built on AWS, and uses Lambda, Cloudformation, ECS, DynamoDB, and Cognito, and is mostly written in Python and Javascript. I've contributed to nearly every facet of development: designing the cloud infrastructure, writing the code for the backend, front end, and our browser extension, as well as engineering the LLM prompts that actually generate the note. I've worked with Claude, Llama-2 and MPT-7B/30B. I also regularly collect and integrate feedback from the physical therapists that use our software.

A major area of study in immunology is the specificity of MHC-I alleles. In my research, I worked to combine machine learning tools that predict MHC-I-Peptide binding with mass spectrometry database search tools, which match the spectral data generated by LC-MS/MS experiments with peptides in the relevant proteome. This culminated in the masters thesis: "

Using Machine Learning to Improve the Detection of MHC-I Presented Peptides" (which can be viewed at: https://ctdigitalarchive.org/islandora/object/20002%3A860706848)

Teaching assistant for Systems Programming, a course about C programming and Linux. Taught lab sections and did grading.

TA'd Systems Programming, a course about Linux and C programming. Taught two lab sections and graded.

Wrote software in Python to parse weblogs, and extract key information, to run on AWS lambda. It would detect which column held date/time of the request, the source IP address, and the target. It would detect the date/time format, and even detect if this format changed during the course of logging. Wrote performance critical sections in C++.