At Fresenius Medical Care, I built and deployed a RAG agent as its sole developer, combining keyword and vector search across internal wiki and repository pages. The department uses its source-cited answers daily.
I also built Databricks pipelines for healthcare reporting and moved a device-history report to automated weekly reporting. I engineered longitudinal patient features and repeatable pipelines for an XGBoost model that identified patients at risk of leaving care.
At Aptiv, I rebuilt and tuned a radar-perception validation stage, reducing false positives by approximately 25% in end-to-end validation. I also developed components that transformed raw radar data into model inputs and standardized formats across three global teams.
At Rochester Institute of Technology, I’m researching synthetic training data for mouse-dynamics authentication as a first author. My projects include Maarg, a claims-support prototype, and a PyTorch model for predicting gene responses to unseen drug compounds.
