● Accelerated data migration of digital library records and unstructured data from CSV, Json and XML using Python (Pandas, NumPy, SQLite), migrating 40% more records in 75% less time than expected. ● Identified data patterns and trends; built Python pipelines for bulk imports, reducing manual effort by 25%. ● Applied rule-based validation and cleaning with Python and Excel to improve data accuracy and reduce cataloging errors.

● Built an LLM-powered pipeline to scan and filter 10,000+ research papers, achieving 85% accuracy in surfacing high-quality studies using transformer models (BERT, RoBERTa, GPT), analogous to object detection in vision tasks. ● Fine-tuned models and optimized data ingestion workflows, boosting theme classification accuracy by 30% and reducing runtime by 40%, showcasing scalable deep learning for high-throughput NLP pipelines.

● Developed ML models (Random Forest, XGBoost, LSTM) for supplier trend forecasting and risk classification, boosting accuracy by 20% and accelerating issue resolution by 35%. ● Built automated data pipelines using Python and Airflow to deliver clean, timely training data from drop tests and quality logs, improving pipeline reliability by 40%. ● Integrated ML outputs into interactive Tableau dashboards to visualize failure patterns and risk scores, enabling QA teams to prioritize high-impact supplier issues. ● Enhanced Power Apps-based Quality Lab system by integrating bulk upload and new test configurations, streamlining global lab processes and reducing manual data entry by 60%.

● Contributed to fraud detection and revenue forecasting models using PySpark, Scikit-learn, and XGBoost, improving signal accuracy by 20% and enabling data-driven financial planning. ● Managed scalable ML pipelines on AWS using Airflow, Redshift, S3, and containerized PySpark, ensuring reliable training and inference workflows. ● Supported real-time fraud detection via ML-driven anomaly pipelines using Kafka, AWS Lambda, and Tableau, improving risk visibility and reducing detection lag by 15%. ● Streamlined CI/CD workflows for ML model training and deployment using Jenkins, Docker, and Terraform, reducing release time by 65% and improving system reliability.

● Devised a predictive model for order prep time using kitchen load, staff availability, and order history, improving wait time estimates and boosting customer satisfaction by 80%. ● Enhanced kitchen energy efficiency by integrating AdaBoost, CART, and LDA to predict peak operational hours, cutting energy use by 40%, and deploying a real-time monitoring dashboard via Shiny in R.