What Vishnuprakash demonstrated

Built a working rolling-bearing fault detector end to end: extracted physics-based vibration features, identified the fault location from the dominant BPFO envelope peak, trained a shallow decision tree within an embedded budget, and evaluated it on a held-out bearing set with catch-rate and false-alarm metrics. The candidate also justified a 0.40 alarm threshold with concrete maintenance-cost tradeoffs in a clear decision memo.

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