Machine Learning model for predicting credit default risk. Applied LightGBM with SMOTE to resolve class imbalance, boosting minority class recall by 18% with zero data leakage.

Executive Tableau dashboard for profit leakage visibility. Uncovered high-cost logistics in Texas/Ohio and aggressive discounts eroding Furniture margins from 9,900+ sales records.

End-to-end Time-Series Demand Forecasting System for logistics. Custom ContainerFeatureTransformer with cyclic encoding, securing 0% data leakage via Walk-Forward Validation.

Dual Inventory Analysis framework (All vs. Recent 90-day MTA) to identify and filter systemic Ghost Inventory, optimizing automated SURPLUS/DEFICIT logistics signals.

Scraping pipeline with FastAPI and LLMs (Kobolt/Gorq APIs) for automated business news summaries, integrated with BMKG XML weather feeds for vessel risk mitigation.

Designed a 3NF normalized schema across 4 tables and built a Python ETL pipeline to migrate 540,000+ transaction records with optimized batch processing.

Convolutional Neural Network (CNN) architecture to detect pneumonia from chest X-ray scans with 94% accuracy.

Modernized legacy Purchase Order and Work Order systems into a modular React & TypeScript architecture with PostgreSQL backend for sub-second query latency.

Production-grade monitoring using Kolmogorov-Smirnov (KS) tests for live Feature Drift Detection, orchestrating all services via Docker and Nginx reverse proxies.

Deep learning NLP model based on BERT to classify public sentiment on trending financial events with probability distribution output.

Random Forest model achieving 90% accuracy for heart disease prediction, demonstrating proficiency in healthcare ML applications with feature importance analysis.

Quantitative analysis using R on 51 students to study correlation between TikTok usage and academic performance using statistical hypothesis testing.