Disha Mahato completed Machine Learning for Automotive Safety

Built an end-to-end perception validation dossier for a synthetic Aurora-7 detector: defined detection correctly at IoU ≥ 0.5, computed aggregate precision/recall and the VRU false-negative rate, and then sliced recall by lighting, visibility, and class to expose the night and occluded failure modes. The final recommendation was evidence-based and conditional, pairing the worst slices with concrete SOTIF mitigations and linking the analysis back to the real Uber ATG Tempe crash.

• Leakage-Free Detection Definition
• Safety-Critical Metric Selection
• Per-Condition Failure Triage
• SOTIF Hazard Framing