Currently working on: An educator dashboard that takes insights derived from analytics to inform student performance, curiosity, and weak spots. A student-facing LLM-enabled prompter that explains key points behind each drawer, supply bin, clinical scenario, or ventilator setting based on student's behavior and performance on website. Informed through a RAG system on a closed-source model - thinking of using Claude, but might also be fun to use Tinker (Thinking Machines) so that I can use cheap PEFT to train models on student performance.