What Gabby demonstrated

Built an end-to-end rolling-bearing fault detector: visualized the vibration snapshot, engineered RMS/kurtosis/crest-factor and envelope-spectrum features, localized the fault to the outer race via BPFO dominance, and trained a compact decision tree within an embedded budget. Evaluated on a held-out bearing set and defended a cost-based alarm threshold using the candidate’s own catch-rate, false-alarm rate, and maintenance-cost numbers.

• Leakage-Free Feature Engineering
• Honest Model Evaluation
• Physics-Based Fault Localization
• Cost-Based Decision Making