AI Safety & Agent EvaluationIntermediateAvailable
Building Evaluation Datasets that Reflect Reality
Tuesday, June 9, 2026 · 4:30 PM – 5:30 PM · Hall D1
From production traces to high-signal eval sets: sampling, labeling pipelines, and avoiding eval drift over time.
Topic
Eval Datasets
Time of day
Afternoon
Speaker
Ingrid Sørensen
Company
Polaris Research
Room
Hall D1
Session ID
S018