Jaya Harish Rajendran completed Machine Learning for Automotive Safety

Built a perception-validation dossier that correctly defined detection at IoU >= 0.5, scored a synthetic detections table, and separated headline aggregate performance from the safety-critical VRU false-negative rate. The work also identified the dangerous night and occluded slices, ranked failure modes by risk, and framed them as SOTIF functional insufficiencies with concrete mitigations.

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