What Kashish demonstrated

Built a perception validation dossier end to end for a synthetic Aurora-7 detector: defined detection correctly at IoU ≥ 0.5, computed aggregate and VRU safety metrics, and sliced recall by lighting, visibility, and class rarity to expose the dangerous night and occlusion failures. The final recommendation was evidence-based and internally consistent, tying the release decision to the measured VRU miss rate, worst-case recall, and concrete SOTIF mitigations.

• Leakage-Free Detection Definition
• Honest Safety Metric Evaluation
• Per-Condition Failure Triage
• SOTIF Functional-Insufficiency Reasoning