Data Scientist at Northern Trust

Developed a PyTorch-based NLP model for sentiment analysis, achieving approximately 30% accuracy improvement and enabling data-driven customer experience strategies. Leveraged large language models to process unstructured text, extracted insights and identified key trends. Implemented named entity recognition (NER) to automate information extraction from legal documents, improving processing efficiency by ~40%. Built Generative AI techniques to address data-scarcity challenges in churn prediction, enhancing model robustness and predictive accuracy by ~35%. Executed LangChain pipelines for real-time knowledge retrieval from structured and unstructured sources, improving data access efficiency for analytics teams. Used Scikit-learn to train and evaluate multiple models, accelerating development by ~45%. Deployed optimized data storage and retrieval using AWS S3 lifecycle policies and Redshift Spectrum, reducing long-term storage costs by ~40%.

Teaching Assistant – Data Science & ML at DePaul University

Led data analysis and predictive modeling initiatives using Python (scikit-learn, NumPy, Pandas) and Spark (MLlib, PySpark), improving lifetime value prediction accuracy by 25%. Built serverless data processing workflows with AWS Lambda, integrating with S3, DynamoDB, and Redshift, reducing pipeline execution time by 40%. Established interactive Power BI reports with advanced DAX and data modeling techniques, enhancing decision-making efficiency by 20%. Performed PySpark DataFrame and SQL transformations to optimize large-scale data processing, boosting query performance by 35%. Supported hyperparameter tuning and A/B testing for marketing strategies to optimize ROI.

Teaching Assistant - Data Science & ML at DePaul University

Automated 10+ recurring datasets, built interactive Power BI dashboards for Advising, Registrar, Financial Aid, and Student Success; processed 500GB+ of student records; tuned XGBoost models to predict at-risk students (60% → 75% accuracy); conducted A/B tests improving engagement by 15–20%.

Data Science Engineer at Hexaware Techno

Engineered an intelligent user analytics and recommendation platform using Python, Pandas, and AWS, enabling personalized content delivery and increasing engagement by 32%. Designed a collaborative-filtering-based recommendation engine and predictive modeling to improve content discovery and satisfaction by 27%. Architected real-time ETL pipelines using AWS Glue, S3, and Lambda for scalable ingestion and processing. Developed predictive churn models to identify at-risk users and support retention strategies. Automated ML workflows and integrated models into production with robust monitoring. Improved dashboards in Tableau and Power BI for real-time insights into user behavior and feature adoption. Streamlined cloud-based ML pipelines in AWS SageMaker for training and deployment.