AI Document ETL Pipeline
SourcePython, FastAPI, PostgreSQL, Gemini, Tesseract OCR, Streamlit, Docker
- Engineered an end-to-end ETL pipeline that ingests unstructured PDFs and images and turns them into structured, queryable records.
- Integrated the Gemini LLM with Tesseract OCR to automate field extraction, cutting manual data entry out of the loop.
- Designed a multi-tenant PostgreSQL architecture with isolated per-user storage behind secured REST endpoints.
- Built a Streamlit monitoring dashboard surfacing real-time pipeline metrics; the whole stack ships as one Docker Compose file.