Priya Ramanathan

Verify model adherence to specified output formats.

Test model resilience against prompt injection attacks.

Compare performance between different versions of a model.

Test model behavior on unusual or boundary inputs.

Evaluate how faithfully a model summarizes content.

Analyze when and why a model refuses to respond.

Assess the quality of model-generated code.

Evaluate model API behavior under rate limiting conditions.

Evaluate model expertise in a specific domain.

Verify model produces consistent outputs for identical inputs.

Assess the logical reasoning capabilities of a model.

Evaluate model performance across different languages.