I transform raw data into business impact.

Problem: Identify fraudulent transactions from highly imbalanced data where fraud cases represent less than 1% of all transactions.

Approach: Applied SMOTE for rebalancing, Isolation Forest for anomaly detection, and built an ensemble (Logistic Regression + Random Forest + Gradient Boosting) with threshold optimization.

Impact: Achieved optimal precision-recall balance, enabling detection of fraud patterns in time and transaction amount distributions.

Problem: AI hiring data is fragmented. Job seekers and companies lack a unified view of salary benchmarks, skill demand, and cross-industry trends.

Approach: Analyzed ~15,000 records across geography, company size, job type, and required skills. Built multi-dimensional Power BI dashboard.

Impact: Revealed AI adoption extending beyond tech into Retail, Media, and Automotive. Identified key market demand drivers for career planning.