Disha Holmukhe

I built Bharat Data Pulse, an end-to-end pipeline for India’s government agricultural market price data. It ingests paginated REST API data and exposes analytics through FastAPI and an interactive Streamlit dashboard.

When I found that the source data’s market, commodity, and date fields didn’t form a unique key, I designed a star schema. I also handled mixed-type prices and day-first date strings with explicit casting.

I fixed an offset-ordering bug that re-fetched duplicate pages and verified the pipeline against API-reported totals. Batched inserts and in-memory deduplication cut bulk load time from a five-hour timeout to under 10 minutes.

I scheduled daily incremental loads with GitHub Actions, including a watermark and an API health check. As a Research intern at RCIIF, I analyzed data-mapping workflows and practices to support DPDPA compliance.