Things I've built. Click any card for the full picture.
ML Diagnostics Engine
Catches data problems before they reach the model. Class imbalance, silent leakage, distribution drift. Runs a full battery of statistical checks on any tabular dataset and returns a structured risk report before training begins.
Sleep Breathing Irregularity Detection
Detects abnormal breathing events during sleep from multi-channel physiological signals. A 1D CNN classifies 30-second signal windows as normal or abnormal, evaluated with leave-one-participant-out cross-validation for subject-independent generalization.