The problem
Quantum concepts are abstract, while generic AI tutors can produce fluent explanations without checking whether their reasoning matches the underlying system.
The approach
Pair an agentic teaching workflow with executable quantum simulations, then use the resulting evidence to ground feedback and evaluation.
My contribution
- Designed the learning and agent workflow around inspectable simulation evidence.
- Built the Python research system and TypeScript learning interface as one product loop.
- Developed evaluation paths for feedback quality instead of relying on conversational fluency alone.
What exists now
- A working research platform for testing agent-guided quantum learning.
- A reusable architecture for connecting lessons, tools, feedback, and evaluation.