Overview

AI for Science is a developing research interest that connects my physics background with tools for more transparent and reproducible analysis. I am particularly interested in agents that execute and verify scientific workflows rather than only generate explanations.

Question

The key problem is how to make AI-assisted research auditable. A useful scientific agent should preserve evidence, expose intermediate decisions, verify outputs, and distinguish calculations from interpretation.

Approach

Potential applications include EELS preprocessing, peak fitting, dispersion tracking, literature retrieval, computational workflow execution, anomaly detection, and structured reporting. This page describes an exploratory direction, not a claim of completed research results.