Applied

Building and validating evaluation dimensions that users can judge reliably in context, and using disagreement as information rather than noise.
Examples: response quality dimensions, judgment stability, trade-offs across dimensions.

Identifying when human judgments are reliable enough to steer system behavior, versus when they should remain advisory or be down-weighted.
Examples: signal strength, missingness patterns, calibration checks, robustness.

Applying research on reputation, social similarity, and selection to inform how platforms design trust signals and reduce unequal outcomes.
Examples: signal design, boundary conditions, unintended consequences.

Translating a dynamic view of trust thresholds, control, and safety infrastructure into concrete system behaviors (e.g., when the system should be cautious, when it can be more autonomous).
This connects to my ongoing model work described on the Research page.

I work best when research questions are tied to a clear decision context (what will change if we learn X?). I prefer simple, testable frameworks; careful measurement; and transparent assumptions. I’m comfortable moving between theory and implementation, and between academic and product audiences.