Eshwar Prasad Y completed Machine Learning for Automotive Safety

Built a perception validation dossier end to end for a synthetic Aurora-7 detector: defined detection using IoU >= 0.5, computed aggregate precision/recall and the VRU false-negative rate, and sliced recall by night, occlusion, and rarity to expose the highest-risk misses. The final ship verdict was evidence-based and internally consistent, with concrete ISO 21448 mitigations and a Tempe crash analysis that mapped the real failure to the dossier’s own night-time pedestrian and rare-class VRU miss modes.

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