November, 2024 – Present
Graduate Research Assistant – Indiana University - Kelley School of Business Indianapolis
- Developed an AI-driven meeting assistant by integrating Zoom’s transcript webhook with a FastAPI service to capture transcript URLs and participant emails in real time.
- Enqueued each transcript event into a Redis-backed RQ queue for asynchronous processing by background workers.
- Implemented a Python worker to fetch the VTT file, generate concise summaries using T5/GPT-4, and email action points via SendGrid, eliminating 80% of manual note-taking and boosting post-meeting productivity by approximately 40%, according to an internal user survey.
- Containerized the entire system with Docker Compose and deployed it on GCP Cloud Run via Artifact Registry for scalable, low-latency performance.
- Built a Python and pandas ETL pipeline to process 1,000+ S&P Global transcripts, classify content, extract metadata, and integrate data into PostgreSQL, reducing manual processing by 50%.
- Extracted financial Q&A data using advanced SQL queries and structured it in Excel (Pivot Tables, VLOOKUP) for GPT-based analysis; improved data accuracy by 20%.
- Automated analysis of 25 communication traits using Python and Excel, uncovering links to stock volatility and reducing analysis time by 30%.
- Created Tableau dashboards with calculated fields, KPIs, and forecasting to support stakeholder reporting and improve turnaround time by 30%.
