Pavneet Thind 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 aggregate performance from the safety-critical VRU metric. The analysis went beyond headline recall by slicing VRU recall across day/night, clear/occluded, and common/rare conditions, exposing a severe night-time and occlusion-driven miss pattern that would matter in release decisions.

• Leakage-Free Detection Definition
• Honest Metric Computation
• Per-Condition Safety Analysis