I built an open-source job tracker that removes manual updates by ingesting job-related emails.

Instead of asking users to log applications or update statuses, the system listens to forwarded emails, extracts structured data using an LLM, and updates the application pipeline automatically.

The interesting parts from an engineering perspective:

- Event-driven email ingestion via webhooks

- LLM-based extraction with deterministic backend decisions - Fuzzy matching to upsert applications instead of creating duplicates - Server-side aggregation and caching instead of heavy DB queries - Fully sandboxed local mode (mock AI, local storage, email trap)

Stack: Java 21, Spring Boot, PostgreSQL, Angular, D3.js.

Demo: https://thughari.github.io/JobTrackerPro

Source: https://github.com/thughari/JobTrackerPro

I’d appreciate feedback on the ingestion design and matching strategy.

Thanks.