Skip to content

Proof of concept

Admissions marksheet and document intake

An admissions office receives thousands of applications in a short window. Mark sheets arrive from different boards and universities, in different languages and layouts, as scans and phone photographs. Staff retype subject wise marks into a spreadsheet, check eligibility by eye, and the compilation takes weeks that the calendar does not really have.

Layers of data being checked before a model is trusted

What we built

  • Intake accepting mark sheets, transfer certificates, category certificates and ID proofs as PDFs, scans and photographs
  • Subject wise extraction of the exact fields the institution uses for eligibility, with each value traced back to where it sits on the page
  • Reading across the major Indian boards and languages, with the values normalised into one comparable format
  • Eligibility rules applied automatically, so applicants are tagged against the criteria rather than sorted by hand
  • A review queue holding only the low confidence pages, with the original document shown beside the extracted figure
  • A merit ready export in the format the admissions committee already works in

What it showed

In the proof of concept clean documents passed straight through into the comparison sheet, and staff spent their time only on the pages that were genuinely unclear rather than reading every application from the start.

What a live engagement needs

A live cycle needs a representative sample of the awkward boards and the worst scans, the current eligibility rules in writing, and one parallel run alongside the manual process so both lists can be compared before anyone relies on the automated one.

Services behind it

Related proof of concept work

See the whole library