I build with LLMs and benchmark them on medical licensing exams — agentically, with Claude Code · Codex · Cursor.
- 🩺 Medicine × AI — evaluating models on real medical exams; deterministic + multi-run variance
- 🧩 Full-stack — Laravel (Prism) · Nuxt / Vue / Vuetify · Python
- 🏠 Own AI infra — self-hosted inference (vLLM), ComfyUI, Proxmox + TrueNAS
- 🔬 Both worlds — frontier APIs + self-hosted open models
