SmartSearch started from a simple, frustrating problem: finding one name in an electoral roll, a scheme list, or a scanned government record, when that name might be spelled five or six different ways across different pages. Names transliterated from Hindi, Marathi, Urdu, and other languages rarely have one fixed English spelling, so a plain text search misses most of what you are actually looking for.
SmartSearch solves that with a phonetic matching engine built from patterns found in real electoral roll data, so searching for one spelling of a name also finds the others automatically. It reads scanned, non searchable PDFs using built in optical character recognition, and it keeps up even with very large documents without slowing down.
Local first, by design
The decision that shapes everything else about SmartSearch is that it never sends your documents anywhere. Every part of the search pipeline, from parsing a PDF to running OCR to matching phonetic spelling variants, runs inside your own browser. There is no account to create and no server holding a copy of what you search. Full detail on exactly what data is involved is in our Privacy Policy.
This was not an afterthought. Electoral rolls and government records contain real, sensitive information about real people. A tool built to search that information carries a responsibility to handle it carefully, and the simplest way to meet that responsibility is to never take custody of the data in the first place.
Who builds it
SmartSearch is built and maintained by Zaid Shaikh. If you have feedback, found an issue, or want to get in touch for any reason, email contact@zaidshaikh.com.