Hi, I'mJan Michalak.

I support medical students and faculty with data science and research at VCOM. Using Python and SQL, I query and analyze large clinical observation datasets from VLMS, VCOM's DoD contract-backed syndromic surveillance platform, built from students' rotation logs. I also build PyTorch and TensorFlow models that surface diagnostic and epidemiological trends for student and faculty research.

I was a full-stack engineer on VLMS, VCOM's Learning Management System. It's a React, Next.js, and MySQL platform that 2,800+ students and faculty use to manage schedules, grades, research, rotation evaluations, and observation logs. Those logged observations feed a DoD contract-backed syndromic surveillance system that does real-time trend detection across VCOM's campuses in the Southeastern U.S. Alongside the product work I handled backend API design, database schema migrations, and dev and production environment config, plus ETL and exploratory data science on incoming data.

Helped build the original PHP, JavaScript, and MySQL version of VLMS, the tool medical students use to log rotation observations. It turned students' field notes into a queryable diagnosis dataset that supports care and research in rural areas. I also kept the production app healthy, triaging user-reported bugs and shipping fixes.

On DARPA's Intrinsic Cognitive Security program, a $8.1M effort led by SRI International, I built LSTM models on AR eye-movement data to flag user vulnerability and simulate mixed-reality "cognitive attacks." Earlier I studied how generative models could exploit weaknesses in iris biometrics on VR and AR headsets, and used LSTM models on gaze data to build adaptive navigation aids.