A hands-on lab showing how “improving” a single metric (AUC/accuracy/F1) can worsen real-world outcomes. Includes metric audits, slice checks, cost-sensitive evaluation, threshold tuning, and decision policies you can defend, so dashboards don’t quietly ship bad decisions.

Advanced Streamlit + Plotly sentiment analysis lab: TF-IDF (word+char), multi-model training, ROC/PR AUC evaluation, cost-aware threshold tuning, error analysis, and live prediction.

IMDB reviews sentiment analysis: EDA → TF-IDF baselines (NB/LogReg/Linear SVM + calibration) → F1 threshold tuning → explainability → BiLSTM baseline.

Decision-first fraud screening: Streamlit analytics UI + FastAPI inference, artifact-locked RF/XGB models, and threshold policy controls.

SMS spam detection pipeline: dual TF-IDF (word+char) → calibrated Linear SVM, nested CV + threshold tuning (F1) + explainability + robustness tests.

Decision-grade donor outreach policy: calibrated P(donated_next_6m) scoring + budgeted Top-K actions + net-benefit optimal threshold, with exported deployable artifacts.

Digital habits → wellbeing risk (binary): calibrated risk scoring + cost-aware thresholding with deployable artifacts (LogReg/RF/XGBoost).

12h clinical deterioration early-warning baseline: tabular models (LogReg / HistGradientBoosting / optional XGBoost), simple probability ensemble, and cost-based threshold/policy tables exported for dashboards.

Machine learning fraud detection system using tabular transaction data. Explores class imbalance, precision-recall tradeoffs, and threshold tuning to optimize real-world performance.

End-to-end ML pipeline for UCI Heart Disease classification. Includes leak-safe preprocessing, baseline + Random Forest + HistGradientBoosting, val-tuned thresholds, and CI that generates a downloadable reports artifact. Best model (HGB) hits F1=0.872, Acc=0.891 on the held-out test set

End-to-end fraud detection pipeline with imbalanced data, probability-based evaluation, threshold tuning, and business-driven model selection using Logistic Regression, Random Forest, and XGBoost.

Fraud detection pipeline with Logistic Regression, Random Forest, and SMOTE — tuned for business trade-offs, evaluated with PR-AUC, precision, and recall.

Cost-aware employee attrition modeling project based on the Salifort Motors case, with operational and survey-rich modes, SHAP explainability, and HR-focused retention insights.

Semiconductor pass/fail prediction project using the UCI SECOM dataset with leakage-safe preprocessing, model comparison, and threshold-based evaluation.

Fraud detection using Random Forest with precision-recall evaluation and threshold tuning on imbalanced data.

Worked on a Fraud detection supervised machine learning case study and built a fraud detection end-to-end model along with a risk framework thhrough root cause analysis and cost-benefit threshold tuning to improve fraud identification and decisioning.

Anomaly detection framework for distributed systems reliability.

Identifying rare event.

Portfolio case study from EAI6020 using Vertex AI AutoML for credit card fraud detection, with precision-recall evaluation, threshold selection, and business trade-off analysis.

This is an end to end machine learning project using my personal shopping data collected over the past three years.