Harsh Mehta

Built a modular AI pipeline combining Tesseract OCR and PyMuPDF to extract data from mortgage documents >200 pages and optimized a LlamaIndex-based RAG system to improve document retrieval precision. Established benchmarking for OCR and RAG performance and produced a technical report and stakeholder UI summarizing model trade-offs.

Selected by NSF I-Corps to validate Ensemble's product-market fit, conducting 90+ customer interviews to refine roadmap and strategy. Built autonomous AI agents and real-time scheduling via n8n and OAuth/PostgreSQL integrations, driving a 35% uplift in booking conversions for the MVP.

Built an end-to-end data pipeline with AWS S3, Glue, and Redshift to centralize financial data and reduced processing time by 63%. Trained and deployed a SageMaker Canvas model achieving 99.19% accuracy and 0.981 AUC-ROC to predict bankruptcy risk and informed investment recommendations.

Evaluated 30+ startups and processed 50,000+ data points using Python to extract KPIs and inform investment recommendations. Built a lightweight ETL pipeline in Snowflake that improved data integration efficiency by 32% and wrote optimized SQL reducing key query times by 45%.

Analyzed 11M+ parking transactions to identify usage and weather dependencies, reducing campus parking search time by 27% and improving stakeholder understanding via interactive dashboards. Applied clustering and predictive models to segment facilities and improve resource allocation by 18%.

Conducted competitive and consumer research that informed a revised pricing strategy increasing sales volume by 10% and a marketing approach that boosted engagement by 15%. Applied Python NLP to 2,000+ reviews and produced Tableau/Power BI visualizations to guide campaign targeting, improving targeting metrics by 11%.

Improved operational efficiency by 35% using analytics and predictive models, built a financial forecasting model that supported strategic decisions and increased revenue by 10%. Identified high-value customer segments that raised lifetime value by 15% and retention by 10% while managing Oracle databases and ETL workflows.

Led data-driven media strategy that increased customer satisfaction by 30% and improved lead conversions by 21%; optimized digital campaigns to raise CTR by 40% and cut acquisition costs by 20%. Used Google Analytics and BI tools to identify high-conversion opportunities, boosting conversions by 25% and ROI by 10% in three months.

Master of Science in Information with emphasis on data, machine learning, and cloud technologies; completed coursework and applied projects from September 2023 to May 2025.

Completed a Bachelor of Management Studies covering core business and management coursework from June 2013 to July 2017.