Pratik Singh completed Machine Learning for Automotive Safety

Built a complete perception-validation dossier for a synthetic Aurora-7 detector: defined detection with IoU≥0.5, scored aggregate and VRU precision/recall, and exposed a severe night/occlusion collapse in vulnerable-road-user recall. The final recommendation was evidence-based and internally consistent, pairing a GO-WITH-CONDITIONS verdict with concrete operational restrictions and SOTIF mitigations.

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