Designed and executed large-scale data cleansing workflows on 10 TB+ legal data across formats (emails, PDFs, chats), reducing irrelevant records by 50% and speeding up document review by 40%. Developed and maintained automated data pipelines using SQL queries and Pandas scripts to extract, transform, and validate metadata (custodian, file types), boosting processing accuracy by 98% and reducing manual effort. Deployed processed datasets to AWS S3 and Relativity One, enabling secure, indexed access to 20+ stakeholders and improving legal document retrieval efficiency by 35%. Managed the complete eDiscovery process from data ingestion to export, writing case-specific search queries and validation checks that ensured 99% accuracy in retrieving legally relevant documents on deadline.

Data Science Intern at DC Ikigai Private Limited

December 1, 2021 - October 16, 2025

Engineered and tested data models for 50+ financial indicators, using SQL and Pandas to speed up portfolio metric calculations by 30%, helping analysts make faster investment decisions. Built a real-time stock prediction pipeline using regression algorithms in Scikit-learn (Random Forest, Linear), applying feature selection and model evaluation techniques to reach 76% accuracy on 1,800+ stock records. Integrated Flask-based APIs with the Ikigai Alpha trading dashboard, automating live predictions and personalized alerts, which cut manual stock lookup by 40% for users.

Machine Learning Intern at Zebo.AI

June 1, 2021 - October 16, 2025

Implemented a fine-tuned EfficientNet-based CNN on 15K+ annotated facial images to classify aging signs, achieving 83.2% accuracy through data augmentation, normalization, and class balancing. Created visual performance reports using Matplotlib and Pandas, tracking ROC-AUC, precision-recall, and error patterns, which improved minority class recall by 21% and reduced false positives by 19%.