Telangana Electoral Roll Search Engine
Java 21 · Spring Boot · React / TypeScript · Oracle ATP 19c · Railway
A search engine over 1.23 crore voter records across 6 districts, built to fix a real problem: the Election Commission's PDFs use corrupted Type3 font encoding that breaks Telugu text search entirely.
- Built a data ingestion pipeline with Tabula-java and Apache PDFBox to extract, normalize, and load records at scale from multi-format ECI PDFs.
- Designed a Telugu consonant-skeleton normalization algorithm that strips diacritics to canonical forms, enabling fuzzy search across variant spellings — the actual fix for the encoding corruption.
- Built a thread-safe immutable in-memory index via Spring Boot
@PostConstructfor concurrent reads with no synchronization overhead.