At Think360.ai, I built a credit risk and fraud detection model from scratch, achieving 93–95% accuracy while researching and validating 6–7 million financial transactions weekly. I also developed research frameworks for banking clients and evaluated model performance using precision, recall, and F1-score.
I lead client due-diligence proof-of-concepts and work with clients and internal stakeholders to deliver analysis, reports, and models that meet their requirements. My work includes Python ETL pipelines, PySpark processing, and data governance across financial datasets.
At Aakash Powertech Pvt Ltd, I developed a model to extract and verify data from Sales Order PDFs, reducing manual processing time by up to 50% and eliminating data entry errors. I also built Power BI and Looker dashboards and automated research and reporting workflows.
Earlier at Think360.ai, I worked on PDF and image-based financial document annotation, OCR, and dataset quality. I improved ML model accuracy by 30%+ through research datasets and classification, and reduced dataset errors by 50% with quality checks and anomaly detection.
